Está en la página 1de 222


Microstructures and Properties

Geoffrey Parrish
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬

Preface to First Edition........................................................................v

Preface to Second Edition .................................................................vii
Introduction and Perspectives ............................................................1
Why Carburize Case-Harden? ........................................................................1
Variability ........................................................................................................3
Laboratory Tests ..............................................................................................3
Design Aspects ................................................................................................4
Case-Depth Specifications ..............................................................................7
Chapter 1: Internal Oxidation..........................................................11
Factors Promoting Internal Oxidation .........................................................11
The Internal Oxidation Process ....................................................................13
Effect on Local Microstructure ....................................................................18
Influence on Material Properties ..................................................................23
Measures to Eliminate or Reduce Internal Oxidation.................................30
Summary ........................................................................................................33
Internal Oxidation.......................................................................................33
High-Temperature Transformation Products............................................33
Chapter 2: Decarburization ..............................................................37
Decarburization Processes ............................................................................37
Influence on Material Properties ..................................................................43
Control of Decarburization ...........................................................................47
Summary ........................................................................................................47
Chapter 3: Carbides ...........................................................................51
Chemical Composition..................................................................................51
Massive, Network, and Dispersed Carbides ................................................53
The Formation of Carbides ...........................................................................60
The Effect of Network and Dispersed Carbides on Properties ..................62
Globular Carbide Dispersions and Film Carbides.......................................69
Globular Carbides and Heavy Dispersions ..............................................69
Film Carbides..............................................................................................70
The Effect of Globular and Film Carbides on Properties ........................70
Summary ........................................................................................................73
Chapter 4: Retained Austenite .........................................................77
Austenite Formation......................................................................................77
Austenite in the Microstructure....................................................................81
Effect on Material Properties........................................................................81

‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Control of Retained Austenite ......................................................................93
Summary ........................................................................................................94

Chapter 5: Influential Microstructural Features ..........................99

Grain Size ......................................................................................................99
Evaluation of Grain Size..........................................................................100
Effect of Grain Size on Properties ..........................................................104
Factors Influencing Microcracking .........................................................108
Formation of Microsegregation ...............................................................113
Effects of Microsegregation on Properties .............................................117
Nonmetallic Inclusions ...............................................................................119
Origin of Nonmetallic Inclusions ............................................................119
Effects of Nonmetallic Inclusions ...........................................................121
Consequences of Producing Clean Steels ...............................................128
Summary ......................................................................................................129
Grain size ..................................................................................................129
Microcracks ..............................................................................................129
Nonmetallic Inclusions ............................................................................130

Chapter 6: Core Properties and Case Depth................................135

Core Factors.................................................................................................135
Core Hardenability ...................................................................................135
Core Microstructure and Hardness..........................................................140
Core Tensile Properties............................................................................140
Core Toughness ........................................................................................143
Effects of Core Properties........................................................................145
Case Factors.................................................................................................148
Case Hardenability ...................................................................................149
Case Carbon Content................................................................................150
Case Depth ................................................................................................155
Quenching Methods ....................................................................................164
Distortion .....................................................................................................164
Summary ......................................................................................................165
Core Properties .........................................................................................165
Case Depth ................................................................................................167
Case Carbon ..............................................................................................168

Chapter 7: Postcarburizing Thermal Treatments.......................171

Tempering ....................................................................................................171
Tempering Reactions ...............................................................................171
Effects of Tempering................................................................................175
Additional Process Factors ......................................................................183
Summary ......................................................................................................194
Tempering .................................................................................................194
Refrigeration .............................................................................................195

Chapter 8: Postcarburizing Mechanical Treatments .................199

Grinding .......................................................................................................199
Grinding Action........................................................................................199

‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Grinding Burns and Cracks......................................................................200
Effect of Grinding Variables ...................................................................203
Residual Stresses Caused by Grinding ...................................................207
Effect of Grinding on Fatigue Strength ..................................................208
Roller Burnishing ........................................................................................212
Effect on Microstructure ..........................................................................212
Effects on Material Properties .................................................................214
Shot Peening ................................................................................................216
Process Control.........................................................................................216
Effect on Microstructures ........................................................................217
Effects on Material Properties .................................................................218
Summary ......................................................................................................222
Grinding ....................................................................................................222
Shot Peening .............................................................................................223
Index ...................................................................................................227

‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
ASM International is the society for materials engineers and scientists, a
worldwide network dedicated to advancing industry, technology, and
applications of metals and materials.

ASM International, Materials Park, Ohio, USA

This publication is copyright © ASM International®. All rights reserved.

Publication title Product code

Carburizing: Microstructures and Properties 06677G

To order products from ASM International:

Online Visit

Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Fax 1-440-338-4634

Customer Service, ASM International

9639 Kinsman Rd, Materials Park, Ohio 44073, USA


American Technical Publishers Ltd.

27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX, United Kingdom
In Europe
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)

Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550

Terms of Use. This publication is being made available in PDF format as a benefit to members and customers of ASM
International. You may download and print a copy of this publication for your personal use only. Other use and distribution is
prohibited without the express written permission of ASM International.

No warranties, express or implied, including, without limitation, warranties of merchantability or fitness for a particular
purpose, are given in connection with this publication. Although this information is believed to be accurate by ASM, ASM
cannot guarantee that favorable results will be obtained from the use of this publication alone. This publication is intended for
use by persons having technical skill, at their sole discretion and risk. Since the conditions of product or material use are
outside of ASM's control, ASM assumes no liability or obligation in connection with any use of this information. As with any
material, evaluation of the material under end-use conditions prior to specification is essential. Therefore, specific testing
under actual conditions is recommended.

Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale, use, or reproduction, in
connection with any method, process, apparatus, product, composition, or system, whether or not covered by letters patent,
copyright, or trademark, and nothing contained in this publication shall be construed as a defense against any alleged
infringement of letters patent, copyright, or trademark, or as a defense against liability for such infringement.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Carburizing: Microstructures and Properties Copyright © 1999 ASM International®
Geoffrey Parrish, p1-9 All rights reserved.
DOI: 10.1361/cmap1999p001

Introduction and Perspectives

Carbon case hardening, through natural evolu- tered by subzero treatment after quenching. Cold
tion, commercialism, and economics, has be- working by either peening or rolling can modify
come a process for which the possible number of the surface microstructures and have significant
variables is so large that it is hardly likely that bearing on the life of the component, as too can
any two companies will process exactly the same. surface grinding.
There will always be some difference in choice One must not overlook the value of the
of materials, equipment, or technique, and there microstructure and properties of the core or of
will often be differences in the quality of the the influence of inherent features such as
product. There may even be conflict of opinion microsegregation, cleanliness, and grain size.
regarding what is good practice and what is bad, The aforementioned structural variants are the
and what is a valid test and what is meaningless. subject of this review, and where possible, exam-
For each component treated, there is an optimum ples of their effect in terms of properties are
material and process combination, but who given. Those properties mainly referred to are
knows what this is for any given component? bending-fatigue strength, contact-fatigue resis-
Most conflicts stem from there being too great a tance, hardness, and wear resistance. These
choice of materials or process variables and from properties were chosen because it is to promote
the wide range of components that are required one or more of these properties that the carburiz-
to be case hardened. ing treatment is employed. A gear tooth is a good
Despite all this, what the carburizing processes example in which each of these must be consid-
have in common is that they produce at the sur- ered. Some significance has been placed on the
face of the component a layer of carbon-rich residual stresses developed during carburizing
material that after quenching, by whichever because these are additive to the applied stresses.
technique, should provide a surface that is hard.
Regrettably, this is no indication that the case-
hardening process has been successful. Additional
microstructural features may exist along with, or Why Carburize Case-Harden?
instead of, the aimed-for martensite, and these in-
deed can significantly influence the properties of With some through-hardening steels, it is pos-
the component, thereby affecting its service life. sible to develop hardnesses equal to the surface
The microstructural features referred to are in- hardnesses typical of case-hardening parts; how-
ternal oxidation, decarburization, free carbides, ever, machine parts (for example, gears) would
retained austenite, and microcracks in the not be able to transmit as much load as would
martensite. case-hardened parts. This is because case hard-
Further modifications to the martensite in par- ening produces significant compressive-residual
ticular can be effected by tempering, and the pro- stresses at the surface and within the hard case,
portions of austenite and martensite can be al- whereas with through hardening, the residual
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
2 I Carburizing: Microstructures andProperties

stresses are much less predictable. Furthermore, shallow (0.3 to 0.6 mm, effective), even with long
high-hardness through-hardened steels tend to processing times, for example, 80 hours. The shal-
lack toughness; therefore, in general , through- lowness of the case limits the range of application
hardened and tempered steels are limited to of nitrided steels. For gears, the limiting tooth
about 40 HRC to develop their best strength-to- size is about 2 rom module (12.7 dp) without
toughness properties. To produce compressive- downgrading. However, within its safe range of
residual stresses to a reasonable depth in a application, the case shallowness provides good
through-hardening steel, one must resort to a lo- bending fatigue, contact fatigue, wear, and scuff-
cal thermal hardening process, such as induction ing resistance.
hardening, or an alternative chemicothennal treat- Carbon case hardening can be employed to
ment, such as nitriding. achieve a wide range of effective case depths (up
When induction hardening is used for gears, to greater than 4 rom) in a wide range of steels
for example, the preferred hardness distribution (limiting core carbon is normally 0.25%) with
is generally to have about 55 HRC at the surface surface carbon contents of approximately 0.9%
and 30 HRC in the core (Ref 1); consequently, and hardnesses of about 60 HRC . The contact-
parts so treated do not have a contact strength or fatigue and bending-fatigue strengths are regarded
wear resistance that are quite as good as in car- as superior to induction-hardened surfaces and to
boozed and hardened parts. The induction hard- nitride-hardened surfaces (above a certain size
ening process is useful for large parts that need limit). The drawbacks with carbon case harden-
to be surface hardened but would distort or grow ing are distortion, growth, and costs. Distortion
excessively if carburized and hardened. Typical and growth are controlled as much as possible
gear steels surface hardened by induction are during heat treating (by the use of dies and plugs)
4140 and 4340 (initially in the hardened and and finally corrected by a limited amount of grind-
tempered condition), and typical case depths ing. The costs are justified in the product to ob-
range from 1.0 to 3.0 rom. tain a high power-to-weight ratio and durability.
Nitriding is a means of producing a hard sur- An indication of the advantages of case hard-
face with high surface compressive-residual ening, compared with through hardening, is
stresses. It is a subcritical temperature process, shown in the torque-speed plots of Fig. 1 (Ref 2).
and consequently, it is an essentially distortion- Here, the safe operating zone for case-hardened
and growth-free process. The degree of hardening gear sets is much greater than it is for through-
relates rnainly to the chromium content of the hardened steels. This means that to transmit the
steel so that a carbon steel will nitride harden same power at a given speed, a set of case-hardened
only a little . Steel 4140 will harden to about 600 gears can be significantly smaller and/or lighter
to 650 HV, and a 3% Cr-Mo-V steel will than a set of through-hardened gears. Alterna-
achieve more than 800 HV. Unfortunately, the tively, size for size, the case-hardened gear set
cases that can be achieved due to nitriding are will be much more durable.

Precision hobbed shaved

gears. ground gears. all gears
without involute correction
a; 3.5 r----,.----.--~--r---__,
Umited by scuffing (precision
Q) 3 hobbed or shaved gears) - + - -- t- - - - l
Urril'ed by wear
"0 2.5 t-- --+F---r----..!...---'--+-- - t -- ---1 "0 2.5 1---+----'''''1--=-""'''''~~_t.
~ Good quality commerical hobbing ~
2 p....,------1i~ -+-- or .lshaPing 'u
- Urn ted tooth fracture
~ 1.5 uffiiiiid Pitting ~ 1.5 H----+---+--'
:> Urriled by
se- l l--J~::.......:l"""'; :;::-- 7 ' f---'--- +-- - j
Urn ted by s
E! scuffing

.~ scuffing .~ 0.5 t-- --+- - -+-- - +-- - t-----1

]i Cil
a: a; OL..-----'--:---'-:~-""--:---'_:_-~
lOS a: 10 lOS l OS
Pilch line velocity, ftImin

(a) (b)
Fig. 1 Failure regions of industrial and automotive spur and helical gears. (a) Through hardened, 18().-350
HB. (b) Precision gears, surface hardened
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Introduction and Perspectives / 3

Variability grade and fastidious for the precision gear. In

most cases, the heat-treatment procedures will be
to the same standard, and the heat treater will
Over the past several decades, the steelmaking perform in the best way possible every time.
industry has moved from basic open-hearth steel
manufacturing to processes such as VIMNAR;
consequently, the quality and consistency of
steels have improved appreciably. Heat-treatment
furnaces have improved, as have atmosphere and
Laboratory Tests
temperature control systems. Additionally, the
gas-metal reactions, carbon diffusion, and other Laboratory tests to determine the effect of
processes that take place during the carburizing metallurgical variables, for example, carbides,
and hardening of steels have become much better retained austenite, and core strength, are very
understood. Add to these factors the introduc- useful and have contributed appreciably to the
tion of quality systems that favor process and understanding of the influences of metallurgical
product consistency, and, all in all, there has features on material properties. However, there
been considerable improvement (a far cry from are problems associated with laboratory testing
the days of pack carburizing). Having said that, that must be recognized and, where possible, al-
absolute precision is not attained because, lowed for. One problem is that the test specimen
among other reasons, exact steel compositions and method of loading often bear little relation-
are impossible to achieve, and atmosphere con- ship to the machine part and service conditions
trol during carburizing is, at best, often only able they are supposed to represent. Apart from that,
to produce surface carbon contents of ±O.05%of test pieces are often small in section so that the
the target value. Therefore, some metallurgical proportion of case to core can be high, and the
variability must be tolerated. microstructure can be martensitic throughout the
The grade of steel for a given machine compo- test section. The effect of these factors on the re-
nent design, the carburized case depth, and the sidual stress distribution and on the contribution
target values of surface carbon adopted by a of metallurgical features can limit the value of
manufacturerlheat treater are based on experi- the test findings. Another problem is isolating
ence, design procedures, and guidelines provided the metallurgical feature to be studied; generally,
in national or international standards, and per- when conducting a test to determine the effect of
haps on adjustments indicated by laboratory a process variation or metallurgical feature on
test results. It is difficult to determine the opti- some property, the researcher attempts to isolate
mum metallurgical condition for a given situa- that test subject. Sometimes this is easy, for ex-
tion; what is optimum in terms of surface carbon ample, when determining the effects of temper-
or case depth for a gear tooth fillet is different ing or subzero treatment. Other times, it is not so
from what is optimum for a gear tooth flank. In easy. For example, to determine the influence of
fact, even if the optimum condition is known retained austenite on bending-fatigue strength, a
for any given situation (and this can vary from large batch of test pieces are prepared. Half are
situation to situation), the heat treater probably left as carburized and hardened with a high re-
could not provide it due to the variability de- tained austenite content at the surface; the other
scribed in the previous paragraph and the fact half is refrigerated to transform much of the
that most heat treaters are happy to get surface surface retained austenite. This is a common
hardnesses within a fairly wide 58 to 62 HRC method of arriving at two retained austenite lev-
range, and effective case depths within a 0.25 mm els, but what exactly is being studied? Is it the ef-
range. Further, without considering section size, fect of retained austenite, or is it the effect of
the previously mentioned composition variabil- subzero treatment? It is agreed that there are two
ity could give batch-to-batch core-strength varia- austenite levels. Is it the difference in austenite
tions within a 20 ksi band. Hence, the ideal and levels that causes a difference of fatigue strength,
the achievable are often different. Gear standards or is it the effect of the new martensite and its as-
cater to different classes of gears, and these dif- sociated short-range stresses induced by refriger-
ferent classes require different degrees of dimen- ation that are responsible for the difference? The
sional precision and finish, as well as different manufacture of batches of test pieces that are
standards of inspection. It is unlikely, however, identical apart from the presence or absence of
that the heat treater will be lax for the lowest network carbides is another example. One can
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
4 / Carburizing: Microstructuresand Properties

standardize surface carbon content and vary the tigue) have been derived that are somewhat less
heat treatment, or one can standardize heat treat- than the actual failure values. The basic allowables
ment and vary the surface carbon content. Either are published in the gear standards (e.g.,
way, there will be differences other than the car- ANSIJAGMA 2001 or ISO 6336) (Table la and
bide network. Nevertheless, laboratory testing b). One should consider that for full-scale gear
provides trends and indicates whether a metallur- testing, the metallurgy of the tested gears might
gical feature will have a small or a large effect on be typical of one heat treater's quality, which
the property under study. could rate either high or low against other heat
treaters' qualities. This is another reason for set-
ting the design allowables lower.
Designers also incorporate into a design safety
Design Aspects factors that will account for any adverse effects
of material and manufacturing variability. 'There-
Laboratory test pieces are designed and loaded fore, there are probably numerous case-hardened parts
to fail. Machine parts, on the other hand, are de- performing satisfactorily in service with surface
signed and loaded not to fail. The basic allow- microstructures that contain adverse metallurgi-
able stresses used by gear designers have been cal features. For example, the high-temperature
conservative in order to acknowledge that design transformation products that accompany internal
procedures are not precise enough to cater to the oxidation tend to be frowned upon, yet there are
very wide range of gear designs, and that mate- numerous case-hardened gears in service with un-
rial variability and process variability do exist. ground roots that, therefore, contain degrees of
These basic allowable stresses are derived from internal oxidation. If the test gears from which
actual gear tests and are set at a lower value than the basic allowable stresses were derived had un-
that of failure stress. For example, in Fig. 2, the ground roots and fillets, then internal oxidation
surface-hardened test gears failed due to tooth will be accounted for anyway. A metallurgical
pitting at contact stresses of 1400 to 1500 MFa. feature might indeed lower the strength of a part
These tests represent nitrided marine and indus- (according to laboratory test results), but the ap-
trial gears that have, in this instance, a design plied service stresses must be high enough for
limit of about 1000 MFa (Ref 3). Comparable that feature to be significant and cause failure. If
gear tests have been conducted for case- hard- the basic allowable stress and the gear designer's
ened automotive gears and aerospace gears. safety factor together reduce the service stresses
From these tests, appropriate allowable stress val- to, say, half the failure strength of the part, but
ues (for both bending fatigue and contact fa-

Table 1(a) Basic allowablestress numbersfor

2000 gears, ISO6336·5 1996
CoDl8ctstress Bendingstress
Quality grade Umit(a. l, MPa limit(a. l, MPa IIardDess, HV

1500 Carburized and hardened

~ • AO

Contact stress
ME 1650 525 670-745
~ MQ 1500 452-500(a) 645-745
:i -- .. ---. ---_ .._-- 0·······1········ ML 1300 315 615-800
l!! 1000 Induction hardened

i D
Nilrided gears
r ME
8 D • Failed, full-scale geartest
A Not failed, full-scale geartest
ML 960-1090 225-275 490-655
o Not failed, industrial gears Gas nitrided, through bordened and tempered
(4-15yearsservice) ME 1210 435 500-650
D Not failed, marine gears MQ 1000 360 500-650
(1.5-15yearsservice) ML 785 255 450-650
0.005 0.01 0.015 0.02 Gas nitrided, nitriding steels
Ratio of effectivecase depth (500 HV) ME 1450 470 700-s50
to relative radius of curvature MQ 1250 420 70Q-850
ML 1125 270 65Q-850
Fig. 2 The results of full-scale gear tests (failure by
tooth pitting) and the typical design stresses used for in- Stresses are shown in MPa, and all hardness values are convened to
dustrial and marine gears. DNV, Det Norske Veritas. HV.Designersshouldrefer to the appropriatestandard.(a) Varieswith
Source: Ref 3 core hardness and/orcore strength
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Introductionand Perspectives / 5

the heat treatment has induced a serious adverse ratios and, hopefully, reduced costs. Therefore,
metallurgical feature with a strength reduction the metallurgists and heat treaters must continue
potential of, say, 30%, there still might not be a to contribute to the cause.
problem (Fig. 3). However, if something should Currently, it is believed that the limitations of
go wrong, for example, if a bearing begins to de- the conventional case-hardening steels are fairly
teriorate or the gear is slightly misaligned, in- well understood. Any other gains must be made
creasing the tooth stress, then failure is more through design and process refinements (consis-
likely to occur. tency and accuracy) sufficient to enable revision
It is not suggested here that one should ignore of the design allowables.
the metallurgical condition, or that quality con- The future might never provide a case-
trol should be relaxed because design, to some hardening steel that is superior in all respects to
extent, accommodates metallurgical variability. the conventional grades. Even if it did, the cost
On the contrary. It could be that on many occa- of the steel might limit its use to very specialized
sions the designer's generosity has, in effect, applications. However, it is possible to design a
"saved face" for those responsible for the metal- steel that is superior with respect to one property.
lurgical quality. If the metallurgical variability The newer grades of special-purpose aerospace
could be reduced across the board, and improved gear steels for use at high operating speeds and
quality and quality consistency could be guaran- temperatures exemplify this designing for pur-
teed, then perhaps the basic allowable stresses pose. Examples of such steel are Pyrowear Alloy
could be increased a little. If nothing else, prod- 53 (Carpenter Technology Corp., Wyomissing,
uct reliability would be improved. Designers PAl, CBS-lOOOM VIM-VAR (Tirnken Latrobe
strive to improve their design procedures, manu- Steel Co., Latrobe, PAl, CBS-600 (Tirnken Co.,
facturers aim to produce levels of accuracy and Canton, OH), Vasco X2-M, and Latrobe
finish the designer specifies, and lubrication en- CFSS-42L, for which the steel compositions and
gineers seek to improve their products. Together heat-treatment operations depart sufficiently
these efforts will lead to better power-to-weight from the conventional. Previously, SAE 9310
steel was preferred by the aerospace industry for

Table 1(b) Basic allowablestressnumbersfor

gears,AGMA 2001-C95 160
Quality Contact ....... 6mit Bendiug stress 6mit
grade (SAC).MPa (SAT),MPa Hardness, HV
Carburized and hardened
3 1900 520 650-800
2 1550 450 or 480(a) 650-800 120
I 1240 380 600-800
Induction hardened
100 30% loss of fatigue
,, Iimttdue
2 1310 152 515 minimum ,, 10 an adverse
1345 152 580 ,, metallurgical feature
1172 152 515 ~ ,
1210 152 580 iii
Gas nitrided, hardened and tempered, 4140and 4340 '"
3 1210 460 minimum
1240 485
2 1125 317-372(b) 460
1160 317-372(b) 485
1030 234-276(b) 460
1070 234-276(b) 485
Probable maindesign
Gas nitrided, 2~~ Cr steel 20 I---+--+-----t rangefor infinitelife
3 1300 42O-44O(b) 580 minimum
1490 42Q-44O(b) 690
2 1190 395-4OO(b) 580 O'----'------'-----'-:------'-:-----L-:------'
1350 395-4OO(b) 690 103 1()4 1()5 1()6 10 7 1()ll 10 9
1070 280-31O(b) 580 Stresscycles
1210 280-31O(b) 690
Fig. 3 Theoretically a "safe" gear design can accom-
Stresses are shown in MPa. and all hardness values are converted to modate the presence of an adverse metallurgical feature;
HV. Designers should refer to the appropriate standard. This table is however, there may be other adverse factors involved
for spur and helical gears. (a) Depends on bainite content. (b) Varies
that also erode the difference between the fracture stress
with core hardness and/or strength
and the allowable stress.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
6 I Carburizing: Microstructures and Properties

gears, but its limitations (questionable hot which have a high potential to occur in
strength, for example) inhibited design progress. high-speed, high-temperature gearing (Fig. 1).
The high-temperature limitationsof lubricants for This resistance may, to some extent, make up for
high-speed, high-temperature gearing is another the limitations of the lubrication.
factor to consider. The new grades of steel are Metallurgy is only one factor in a bigger
designed to maintain their strength at operating picture that includes machine and component
temperatures and resist scoring and scuffing, design, manufacturing accuracy, machine

2530 26 367

6/ 0

2431 24 352.5

2327 22
] 6 je 0

2220 20 f J ~

2106 18
/ • // / 305.5

I Iv.'
1985 16 288
'" 8
f ~
l!! 1857
J: 14 (v 269 i
~ V / / -~
J ~
'" 1718
l!! Gl 12 249

// I ~
E ~ 8
8 u,


vo 01 ~/ I

1 VV eJ

J II ,I.·
1216 6 176
•••; ' ~roUgh-hardenedsteels
• 0 Flame-hardened steel
993 4 144
t·· e • Induction-hardened steel

I ~
v Gas-nnrided and san-bath nliridedsteel
• Sulphinuz-treated steel
702 2 ..... o Gas-nlirided (80 h) steel 102
/~ (v). (e) Maraging steel

V 20 40 60 80
• Gas-carburized andhardened
Gas-carburized. hardened andtempered
100 120 140 160 180
Core strength. tsi

Steel Effective casedepth, IDDl (io.)

Through hardened (various)
Flamehardened (PCS)
Induction hardened (4340) 3.75 (0.15)
Gas nitrided and salt-bath nitrided 0.14 (0.005)
Sulphinuz treated 0.17 (0.007)
Gas nitrided (SOh) (3%Cr-Mo) 0.35 (0.015)
Maraging (x) 0.14 (0.005)
Gas carburized, hardened, and tempered (Ni-Cr) 1-1.5 (0_()4.4).06)
Gas carburized and hardened (Ni-Cr) 1.5(0_06)

Fi~. 4 Effect of core strength and case depth on the rolling-contact fatigue limit of gear steels. Tests in-
voTvedtwo 4 in. disks driven by a 2 in. roller. Testpiece may have been either one of the disks or the roller.
Relative radius of curvature, 2/3. SH units = Ib/in. of face width divided by the relative radius of curvature.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Introduction and Perspectives / 7

assembly, lubrication, application, machine use chosen case depth must be adequate to resist the
or abuse, and maintenance (or lack of it). This stress at the dedendum-pitch line area.
book considers some of the current knowledge The contact stress increases with transmitted
regarding the metallurgy of case-hardened steel load so, strictly speaking, the case depth should
parts and what effects or trends the various met- be determined by the load. Using the shear-fatigue
allurgical features have on the properties of such strength (ultimate tensile strength x 0.34) of the
parts. However, it focuses on conventional material as opposed to shear stresses due to load-
case-hardening steels and processing and, there- ing appears to give some conflicting results;
fore, might not be as helpful to designers and us- therefore, it is not clear on which shear stresses
ers of new alloy grades. the case depth requirement should be based. For
example, if the 45° shear stresses ('t yz) are con-
sidered in conjunction with the test results shown
Case-Depth Specifications in Fig. 4, it is found that, for the 80 hour-nitrided
surfaces, the predicted fatigue limit is about half
At the dedendum-pitch line area of a gear of the value determined by testing. On the other
tooth, there is a smaller radius of curvature than hand, the fatigue limits for the carburized, hard-
at locations above the pitch line. Consequently, ened, and tempered surfaces (100 to 200 "C) and
the contact band there tends to be narrower than for induction-hardened surfaces are better pre-
at the addendum so that for a given load, the con- dicted (Fig. 5). The orthogonal shear stresses
tact stresses will be higher. For that reason, the ('tortho)' however, predict fairly well the fatigue

120 r - -....-----,r_--.------r---r----,.--~-__r_--r_-_r_-____,-_____,





20 ~-_+--I-- 0215 ksi core

Nltrkled steels • Maraging
(seealso Fig. 6) A 145 ksi core
.. 110 ksi core
0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 0.110 0.120
Distancefrom surface,in.

Process Predicted roUp_limit, SH Actual roUp_limit, SH

Carboncase hardened,untempered 24,000 18,000
Temperedat 100°C 24,000 22,000
Temperedat 150°C 24,000 23,000
Temperedat 200 °C 22,000 25,000
Temperedat 250°C 20,000 26,800
Inductionhardened 18,000 -18,000
Nilridedsteels Predictionequals aboutone half of actual

Fig. 5 Plots of shear-fatigue strength (from hardness) against plots of shear stresses, "tyz, in rolling-contact
tests. Predicted and actual fatigue limit values are in close agreement for carburized steels but not for the four
nitrided steels. Relative radius of curvature, 2/3. SH units = lb/in. of face width divided by the relative radius of
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
8 / Carburizing: Microstructures and Properties

limits for the nitrided surfaces but overestimate sidual stresses were not measured in either in-
the fatigue limits for the case-hardened and the stance, it is nevertheless likely that the roller
induction-hardened surfaces (Fig. 6). From these tempered at 250 °C had the lowest compres-
apparently conflicting results, it is difficult to sive-residual stress in the case, and the untem-
draw any meaningful conclusions that would pered roller had the highest (see Fig. 7.12). This
help determine the appropriate hardness profile implies that compressive-residual stresses
and case depth for a given application. might not be beneficial where rolling contact is
The relationship of residual stresses to rolling involved-where the fatiguing actions are
contact fatigue is also unclear. The table in Fig. 4 subsurface but still in the case. Therefore, this
shows that for the case-hardened tests, the un- further complicates arriving at a theoretical solu-
tempered roller produced the lowest fatigue tion for determining adequate hardness profiles
limit, and the roller that had been tempered at and case depths. Fortunately, there is still the
250 °C produced the highest value. Although re- well used case depth-to-tooth diametrical pitch
relationship to fall back on, even if it is not
strictly correct (Fig. 7).
Interestingly, with rolling-contact fatigue tests
120,..---.,---.,-----r----r------, of shallow-cased surfaces (i.e., when the depth of
maximum hertzian shear stress is deeper than the
effective case depth), there is no work hardening
100 f---f--+-'" at the case-core junction up to the fatigue limit.
At stresses above the fatigue limit, work harden-
ing does occur, and the extent of the working
(hardness and depth) increases with the contact

hemin=0.119935xPnd .

QC 10
\ NoriiWi

hemin = 0.264693 x
Cl. 7 Pnd-1.124a1
~ 6
W 5 Heavycase depthi
E 1\
0.010 0.020 0.030 0.040 0.050
tii 3
Distance from surface, in. a
z 2

fatigue limit, S8
fatigue limit, SH '\
Carbon case hardened >31,000 20,000-24,000 1 1\ ~
Induction hardened 27,000 18,000 10- 3 2 3 5 10- 2 2 3 5 10- 1 2 3
(a)Nitrided for 80 h, 25,000 25,000
Minimum effective case depth, (he min), in.
215 ksi core
(b)Nitrided for 80 h, 16,500 14,000
maraging steel Fig. 7 Minimum effective case depth for carburized
(c)Nitrided for 80 h, 14,500 15,000 gears, he min' The values and ranges shown on the
145 ksi core case-depth curves arc to be used as guides. For gearing
(d)Nitrided for 80 h, 7,000 7,000-9,000 in which maximum performance is required, detailed
110 ksi core studies must be made of the application, loading, and
manufacturing procedures to obtain desirable gradients
Fig. 6 Plots of shear-fatigue strength against plots of both hardness and internal stress. Furthermore, the
of shear stresses, 'artha' in rolling-contact tests. In con- method of measuring the case, as well as the allowable
trastto Fig. 5, predicted and actual fatigue limit values tolerance in case depth, may be a matter of agreement
are in good agreement for the fournitrided steels but not between the customer and the manufacturer. Effective
the other steels. Relative radius of curvature, 2/3. Shear case depth is defined as depth of case with a minimum
fatigue strength is ultimate tensile strength x 0.34. SH hardness of50 HRC; total case depth to core carbon is ap-
units = lb/in. offace width divided by the relative radius proximately 1.5 x effective case depth. Sec ANSIIAGMA
of curvature. 2001-C 95.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Introduction and Perspectives / 9

Eutectoid Carbon Content martensite and in any retained austenite. Consider

then: is a eutectoid carbon martensite the best to
The requirements and information in any stan- provide all the properties sought for a given appli-
dard are, in general, readily understandable and cation? Or is it the best carbon content for holding
realistic, as they should be. Unfortunately, there the retained austenite to a low value or for develop-
are exceptions. For example, the surface carbon ing a better case toughness? Would a case-hard-
requirement for carburized gears as set out in ened 9310 steel gear with a surface carbon content
ISO 6336-5 1996 is "Eutectoid carbon % of 0.55% be regarded as fit for service even
+0.20%, -D. 1%." The standard does not justify though it might satisfy the case carbon require-
the use of the term eutectoid. It does not provide ments oflSO 6336-5 (1996)?
a list of case-hardening steels along with a repre- To establish where the eutectoid carbon con-
sentative value of eutectoid carbon for each steel, tent figures in deliberation regarding property
nor does it provide an empirical formula for de- optimization for case-hardened parts (and indeed
termining the eutectoid carbon. It is, therefore, it may have a place), there is little alternative but
unhelpful and unworkable as it stands. However, to establish eutectoid carbon data for each steel.
it is understood that the standard is to be revised For this, it may not be necessary to go through
to correct the problem. the complex procedure of determining accurate
The term eutectoid carbon content refers to equilibrium diagrams. Instead, a set procedure
the carbon content that produces only a pearlitic could be devised in which, for example, a 30 mm
matrix microstructure as a result of an extremely bar is carburized to, say, greater than 1% surface
slow cool through the AC3 or ACcm to ACl tem- carbon content and cooled, or heat treated to pre-
perature range. A steel with less than the cipitate the excess carbon as carbides. The bar is
eutectoid carbon content (hypoeutectoid) con- then cut into two: one half is used to determine
tains pearlite with some ferrite, whereas a steel the carbon gradient and the other is used as a
with more carbon than the eutectoid carbon con- metallographic sample to determine the depth of
tent (hypereutectoid) contains some carbide carbide penetration. The two sets of data are then
along with pearlite, again due to very slow cool- brought together to give a value of carbon at
ing. Each steel grade has its own eutectoid car- which, under the set conditions, carbides just ap-
bon content, and considering the whole range of pear. This could then be referred to as the "ap-
conventional case-hardening steels, the eutectoid parent eutectoid." Only with such information
carbon contents could easily vary between 0.45 could the merits of the case carbon requirement
and 0.8%. In case-hardening practice, the cool- of the ISO 6336 standard be assessed.
ing rates employed, even slow cooling from car-
burizing, are much faster than the cooling rates
researchers would use to determine the eutectoid REFERENCES
carbon for an equilibrium diagram. Rapid cool-
ing, typical of commercial quenching, can sup- 1. G. Parrish, D.w. Ingham, and J.M. Chaney, The
Submerged Induction Hardening of Gears, Parts 1
press the formation of ferrite in lean-alloy steels
and 2, HeatTreat. Met., Vol 25 (No.1) 1998, P 1--8,
within about 0.2% C less than the eutectoid and and Vol 25 (No.2), p 43-50
suppress the carbide formation in that steel when 2. M. Jacobson, Gear Design: Lessons from Failures,
the carbon is up to about 0.2% above the Automot. Des.Eng., Aug 1969
eutectoid. Suppression of ferrite or of carbide 3. I.T. Young, The LoadCarrying Capacity ofNitrided
means that the carbon will be in solution in the Gears, BGMA, London, 1982
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Carburizing: Microstructures and Properties Copyright © 1999 ASM International®
Geoffrey Parrish, p 11-36 All rights reserved.
DOI: 10.1361/cmap1999p011

Chapter 1

Internal Oxidation

The presence of internal oxidation at the mosphere has a typical composition of 40% H,
surfaces of parts that are case hardened by 20% CO, 0.46% CH 4, 0.27% CO 2, and 0.77%
pack or gas carburizing has been known of for H20 (vapor; dew point, 4°C), with a balance of
fifty years or more. The high-temperature nitrogen. Such a mixture will have a carbon po-
transformation products (HTTP), which can tential for iron of approximately 0.4% at 925°C;
form as a direct consequence of internal oxi- therefore, in order to effect the carburization of
dation, have subsequently been found to have steel components to the required surface carbon
adverse influences on certain strength proper- levels, endothermic gas must be enriched by
ties of affected parts; therefore, these prod- controlled additions of a suitable hydrocarbon,
ucts are of some concern to metallurgists and such as propane or methane.
engineers. The balance of the component gases ensures
The use of oxygen-free gas-carburizing atmo- that the endothermic atmosphere is reducing to
spheres or vacuum-carburizing processes is iron, the parent metal of the steel, noting that the
known to eliminate the oxidation process, and steel will be in the austenitic state at the tempera-
nitrogen-base atmospheres are said to reduce it. ture for carburizing. However, for those alloying
However, conventional gas carburizing using the elements in solid solution in the steel that have a
endothermic carrier gas is still the most popular greater affinity for oxygen than iron does, the at-
method of case hardening, and its use will con- mosphere is potentially oxidizing.
tinue for many years. Thus, the problems related Elements That Oxidize. Water vapor and
to internal oxidation will persist as long as the carbon dioxide are the offending component
conventional process lives. Therefore, it is gases in the endothermic atmosphere that pro-
important to understand how internal oxidation vide the oxygen for the internal oxidation pro-
comes about, what its likely effects are on mate- cesses. The oxidation potentials of the main ele-
rial properties, and what should be done about it, ments used for alloying can be derived from the
bearing in mind that it has generally been toler- ratios of partial pressures of the oxidizing and re-
ated in the past. ducing constituents in the atmosphere, that is,
pH 20 to pH 2 and pC02 to pCO. The results of
such calculations, as presented by Kozlovskii
Factors Promoting Internal Oxidation et al. in Ref 1 for a temperature of 930°C, are
shown graphically in Fig. 1.1. This diagram
Endothermic Atmosphere. Gas carburizing is shows that of the elements studied, titanium, sili-
normally carried out at a temperature within the con, manganese, and chromium are likely to
range of 900 to 950°C using an endothermic oxidize, whereas iron, tungsten, molybdenum,
carrier gas generated by the controlled combus- nickel, and copper will not oxidize. This, of
tion of another gas (such as natural gas, liquid course, refers to elements that are not combined
propane gas, or towns gas) with air in the pres- (i.e., those in solid solution). 1\\'0 possibly im-
ence of a catalyst at a high temperature. Prepared portant omissions from this diagram are alumi-
from natural gas (methane), the endothermic at- num and vanadium, both of which are common
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
12 / Carburizing: Microstructures and Properties

additions to steels. According to Fig. 1.2 (Ref can be seen that elements favoring internal
2), these elements will oxidize in an endother- oxidation are generally necessary to the steel
mic atmosphere; thermodynamically, alumi- to impart characteristics such as harden-
num appears to be slightly more ready to oxi- ability, toughness, and grain refinement; but
dize than does titanium, whereas vanadium ironically, in some cases, their function is to
will have an oxidation potential somewhere assist in the deoxidization process during
between those of silicon and manganese. It steel melting and casting operations.

12 Calculations based on the equations:

10 Oxidation of
o Me + H20 :? MeO+H2

fd Me + CO2 :? MeO+CO

.e 4
s 2
c:: 0
~ -2 Reduction of
-4 metaloxides

-8'-- Cu

Fig. 1.1 Oxidation potential of alloying elements and iron in

steel heated in endothermic gas with an average composition of
40% H 2 , 20% CO, 1.5% CH 4 , 0.5% CO 2, 0.28% H 20 (Dewpoint,
10 0C), and 37.72% N 2 • Source: Ref 1




1()2 1()2

t'" FelFe304

10- 2




10- 10
600 BOO 1000 1400 600 BOO 1000 1400
Temperature, °C Temperature, °C

(8) (b)

Fig. 1.2 Critical requirements for the oxidation of selected metals with indicated temperatures in atmospheres contain-
ing (a) water vapor and hydrogen and (b) carbon dioxide and carbon monoxide. Source: Ref 2
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxication / 13

The Internal Oxidation Process compounds unless the carbon potential is unduly
high enough to form carbides. When the carbon
Oxygen Penetration. Oxygen is an interstitial potential is high enough to form carbides at the
element in iron, having an atomic size approxi- surface, the oxidation of elements, such as chro-
mately 33% smaller than that of iron (noting that mium, titanium, and manganese, takes place
carbon atoms are -34% smaller than iron atoms). within the carbide particles or at the car-
However, iron has a low solubility for oxygen, bide-austenite interfaces (Ref 5).
and the diffusion oxygen through the ferrous ma- The Depth of Oxidation. In commercial
trix is relatively slow (1<r 9 cm 2/s). As Fig. 1.3 il- case-hardening steels, the depths at which the
lustrates, the depth of total penetration of oxygen oxides are detected by conventional optical mi-
due to 6 hours at 930°C in a typical endothermic croscopy are typically less than 25 um (i.e., for
atmosphere is only about 75 urn. This figure carburized total case depths of 1 to 2 mm).
suggests that from about a 5 urn depth there is a Deeper cases will produce deeper penetrating
steady fall in the oxygen content. On the other oxides; for example, an 8 mm total case depth in
hand, a second high oxygen peak some distance a Cr-Ni-Mo steel would likely have an oxide
from the surface has been observed (Ref 4, 5); penetration depth of75 to 100 urn.
for example, the second peak has been found at As Fig. 1.3 indicates, the depth of oxygen pen-
17.5 urn (Ref 4). etration is much greater than the depth to which
At a given temperature, the oxygen content the oxides form. In this instance, the depth of ox-
~d the depth of oxygen penetration are strongly ygen penetration is about three times as great;
influenced by the oxygen potential of the atmo- the oxides formed in the first 25 urn, (i.e., in that
sphere (the limiting oxygen potential being that layer where there is a high oxygen plateau). If
at which iron begins to oxidize). However, as the oxides had formed at a greater depth, they were
carbon potential rises, the oxygen potential falls; too small to be resolved optically.
consequently, with high-carbon potential carbu- The depth of oxide penetration is influenced
rizing, the oxidizing effect is reduced depending by a number of variables. For example, the depth
on the duration of carburizing. The relationship of oxidation increases with the case depth, and
between time and temperature, with respect to for a given carburizing time, it increases with the
internal oxidation, is shown in Fig. 1.4. temperature of carburizing. The depth of oxida-
In the oxidation process, oxygen atoms, re- tion also increases with a lowering of the carbon
leased by the gas-metal reactions that take place potential (i.e., with an increase of carbon dioxide
during carburizing, are adsorbed onto the metal- and water vapor), and also with an increase of
lic surface. From there, the oxygen atoms diffuse the grain size. A most influential factor is, of
inward along grain and subgrain boundaries and course, the chemical composition of the steel.
into the lattice. Once there, they can chemically Oxide Morphology. Metallographically, two
combine with available substitutional elements oxide morphologies generally form. That oxide
that have a high oxidation potential and form ox- nearest to the surface (typically to a depth of 8
ides. Meanwhile, carbon and hydrogen, the other
interstitials released by the gas-metal reactions,
similarly penetrate the surface to diffuse more
quickly inward because they do not react to form
E ./
E 0.Q16
Jg C V
§ 5
~ 4
~ 3
lA .L

l..-a-' ~O°C
\'I /'

l.L .M
'6 1
§O o
~ 0 10 20 30 40 50 60 70 80 90 100 o 2 4 6 8 10
c( Distance from surface, 11m Carburizing time, h
Fig. 1.3 Qualitative distribution of oxygen in the FiK. 1.4 Depth of the oxidized zones vs.
surface of a Cr-Mn-Ti steel (25KhGT) after carburiz- caiburizing time at different carburizing tem-
ing in an endothermic atmosphere. Source: Ref 3 peratures for SAE 1015. Source: Ref 6
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
14 / Carburizing: Microstructures and Properties

urn) appears as globular particles of about 0.5 The grain size at the surface of the steel is
urn in diameter. This oxide resides mainly in the thought to influence oxide formation in that as
grain and subgrain boundaries and, to a lesser the grain size decreases, the probability of form-
extent, within the grains themselves. Sometimes ing oxides within the grains increases (Ref 3).
it occurs along the surface. Within this surface However, it is suggested that the effect of tem-
zone, the grains are likely to be subdivided into perature on penetration depth, as illustrated in
Fig. lA, might also be affected by grain size
volumes of 0.5 to 1 urn across, although diame-
(Fig. 1.6). One can imagine for a given oxidizing
ters of 2 to 4 um have been quoted (Ref 3). The potential of the atmosphere that the more grain
second type of oxide resides at typical depths of boundaries there are at which to distribute the
5 to 25 um and mainly occupies the prior austen- available oxygen, the less the penetration will be.
ite grain boundaries where it appears as a contin- Thus, steel composition and grain size are in-
uous "dark phase" (Ref 7), dark enough to re- volved in the internal oxidation process. But
semble a void. what about carburizing conditions? What is cer-
One can envisage that, as the oxygen gradient tain is that when carburizing in an endothermic
begins to develop during the carburizing process, atmosphere containing 20% CO and 0.2 to 1%
the globular precipitates will start to form at the CO 2, the format ion of internal oxidation is un-
boundaries nearest to the surface and continue to avoidable. Dawes and Cooksey (Ref 10) est-
grow as adequate quantities of reactants are imate that 0.2% would be the maximum value of
brought together. However, Van Thyne and CO 2 that could be tolerated to prevent the inter-
Krauss (Ref 8) have shown that the formation of nal oxidation of a 1% Cr steel, and that 0.01 %
the globular boundary oxides takes place by a CO 2 would be the limit for a 1% Mn steel.
discontinuous, lamellar growth process ; that is, Mitchell , Cooksey, and Dawes illustrate how in-
rods of the oxide form, each separated from the ternal oxidation increases with manganese con-
next by a band of alloy-depleted austenite. These tent and add that the severity of attack is related
rods tend to grow in the direction of the oxygen to the total case depth (Ref 11). Chatterjee-Fischer
gradient. The oxides appear as rows of spheres agrees with this, stating that samples having
when, in reality, the cross sections of rods are be- comparable case depths, arrived at by carburiz-
ing viewed. At greater depths within the oxi- ing at two entirely different temperatures, would
dized layer, the oxides appear to be continuous have the same depth of internal oxidation, even
and at the prior austenite grain boundaries (Fig . though the morphology of the oxides might dif-
1.5). fer somewhat (Ref 6). For a given temperature,

. . ..
: .... ' -! . -: .
• L.

. ' r: _.

. .

.. ,
Fig. 1.5 Inte rnal oxid ation of a Ni-Cr steel carburized in a laboratory furnace , showing both
grain boundary oxides and oxid e precipitates within gra ins. SSOx
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Postcarburizing Mechanical Treabnents / 15

the increase of depth of internal oxidation is pro- shown in Fig. 1.7 will lead to more of the oxide
portional to the square root of the carburizing being produced. An increase in the silicon con-
time. Edenhofer found that when carburizing a tent, everything else being equal, will influence
16MnCr5 steel with the carbon potential and car- the depth of oxide penetration in a negative way
burizing duration each held essentially constant, and will increase the amount of grain-boundary
doubling the carbon monoxide content from 20 oxide formed. In this respect, it was shown that
to 40% doubles the depth to which internal oxi- increasing the silicon content of the iron-silicon
dation penetrates (Ref 12). alloy from 0 to 1.83% produced isolated fine
The Oxidation of Two-Component Alloys. Where- precipitates to a depth of -20 um when the sili-
as Fig. 1.1 indicates which elements of a steel con was equal to 0.09%; dense globular and
are likely to oxidize during carburizing , it gives grain-boundary oxides were produced to a
no clues regarding how much of anyone ele- depth of only -10 urn when silicon was equal to
ment is needed for the oxidation reaction to 1.83% (Ref 6). Between these two amounts of
take place. Employing pure two-component al- silicon, the quantity of oxide increased with
loys, for example Fe-Si, Fe-Mn, and Fe-Cr, silicon content, but the depth at which it formed
Chatterjee-Fischer confirms that those alloys decreased.
containing elements with a propensity to oxidize The Oxidation of Multicomponent Alloys. With
do indeed oxidize, provided that a sufficient multicomponent alloys (and commercial grades
amount of that element is present (Ref 6). Figure of carburizing steels), the situation is rather more
1.7 summarizes her results and provides infor- complicated. In such alloys the silicon content to
mation regarding atomic number and size. It can cause internal oxidation is about half that for a
be noted that the elements with the larger atomic straight iron-silicon alloy (i.e., 0.05%), which is
sizes, or smaller atomic number, need only be well below the 0.2 to 0.3% silicon content typi-
present in amounts of less than 0.1 vol% to pro- cal of case-hardening steels. This suggests that
mote oxidation, whereas significantly more is while these typical silicon contents are used, in-
needed of those elements whose atoms are of a ternal oxidation during conventional gas carbu-
similar size to that of iron. Alloy contents in rizing will be impossible to prevent. It does
amounts greater than those threshold values seem, however, that by limiting the manganese


11 200


iooo -c



.~ 125
" 1\
/ 930 °C
6 100 \
0/ J 75
V 50
/ 850 °C
r" n " fl;' 'j
4 25

3 o
700 800 900 1000 700 800 900 1000
Temperature. °C Temperature, oC

780 °C
Fig. 1.6 Influence of carburizing temperature on the depth of oxide penetration and penetration fre-
quency per millimeter of grain boundary oxidation for steel 17NiCrMo14. Adapted from Ref 9
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
16 / Carburizing: Microstructures and Properties

and chromium to less than 1% in total, the depth Okasaki, in steels with up to 0.1% titanium, the
of oxide penetration will be essentially that due oxygen, carbon, nitrogen, and sulfur, which in-
to the silicon content. hibit grain growth, combine with titanium to free
The other elements with atomic numbers less the grain boundaries (0.1% titanium having the
than that of iron should be considered. Alumi- maximum effect). More titanium is needed to
num, for deoxidization and grain size control, is repin them (Ref 14). Therefore, a titanium
usually present in soluble form in amounts of ap- content in excess of 0.1 % would be expected to
proximately 0.01 to 0.06%. Though hardly support the internal oxidation reaction. Vana-
enough to cause oxidation in an iron-aluminum dium is not normally added to case-hardening
pure alloy, this could well be enough to do so in grades of steel for the purpose of, for instance,
a commercial grade, according to the behavior of hardenability or strength, for which something in
silicon. To be effective as a grain refiner, tita- excess of 0.1% would be expected. For grain re-
nium must be in excess of 0.1%. According to finement, the amount could be much less than
that, but whether or not it would be below the
threshold for oxidation is not known.

, The Oxidation of Commercial Case-Hardening
Alloys. With commercial grades of steel, the ob-
servations regarding internal oxidation are at
times confusing, which is not too surprising
!\ when one considers that different researchers
0.9 \ have employed different steel compositions and
\Mn carburizing conditions.
~ 0.8 Arkhipov, employing a 18Kh2N4VA steel,
E 0.7 \ found that the silicon and manganese did not ox-

~ 0.6
\ Two component idize, whereas the chromium did (Ref 15). The
ill / chromium content in this case was 1.65%, the
silicon was 0.28%, and the manganese only
0.4 0.4%; with another nickel-chromium steel where
0.3 the chromium content was 0.8%, however, the
\ Multicomponent internal oxidation was less extensive. In yet an-
.) V alliys other study by Arkhipov, this time using a
O.1 v y
, Cr-Mn-Ti steel, the larger oxides observed in the
......0 .....
o 5i Ti AI surface (up to 6-8 urn) contained manganese,
-1Q2 -1Q4 -106 -10B -10 10 -10 12 chromium, and silicon; at greater depths, how-
ever, only the oxides of silicon persisted (Ref 7).
Essentially the same observations were made by
Element Atomic number reJatiwto Iroo, ... Murai et al. (Ref 16) and Chatterjee-Fischer (Ref
Interstitial elements 6). Figure 1.8 shows a typical distribution of ox-
Hydrogen 1 -58
Carbon 6 -34
Nitrogen 7 -36 This reflects on the fact that the composition
Oxygen 8 -33 of an oxide phase at any depth from the surface
Solid-solution elements
Molybdenum 42
is primarily governed by its energy of formation,
Copper 29 1 and the higher this energy is, the deeper the zone
Nickel 28 -1 is in which that particular oxide will form.
Iron 26 o Again, for energy reasons, some oxides tend to
Manganese 25 1
Chromium 24 I form at the grain boundaries while others tend to
Vanadium 23 6 form at sites within the grains, grain size perhaps
Titanium 22 -36
Silicon 14 7
having an influence.
Aluminum 13 14 With regard to the quantity of an oxidizable el-
ement, Mitchell et al. showed in Ref 11 that with
Source: Ref 13
C-Mn-B alloys carburized at 925 EC for 48
hours, internal oxidation was light to a depth of
Fig. 1.7 The limiting amount of added element (of
atomic number less than that of iron) to promote internal -20 urn when the manganese content was 1 to
oxidation. See also Fig. 1.2(bl. Adapted from Ref 6 1.5%. Internal oxidation was heavy and to a
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Postcarburizing Mechanical Treatments /17

depth of -40 urn when the manganese content icon oxide penetrated to a greater depth than did
was 1.5 to 2% the manganese-chromium precipitates. In an-
The Composition of Oxides. The actual compo- other study, the near-surface, grain-boundary
sitions of the oxides have been variously quoted. precipitates were identified as (Crz Mnh04, with
Employing chromium-manganese steels, KaIner SiOx at a greater depth, confined to the fine grain
and Yurasov, who detected oxides to a depth of boundaries (Ref 16). In Ref 19, Preston identi-
30 urn, identified them to be of a complex com- fied chromium, manganese, and vanadium in the
pound that contained at least two metallic ele- internal oxidation of a 0.6% Cr, 0.7% Mn, 0.07%
ments with a spinel structure: nFeOMZ03 (where V steel in which silicon, titanium, and aluminum
M is either manganese or chromium) (Ref 3). It were also present. However, in Ref 20, Dowling
was found that the total M content of the oxide et al. found MnzSi04 in a carburized SAE 4615
was, according to Table 1.1, up to 15% Cr or steel and something akin to MnZCr30g in a car-
11.3% Mn. It was also found that some part of burized SAE 8620 steel.
the M could be replaced by small amounts of ti- Composition Gradients. The metal-oxygen re-
tanium and molybdenum in those steels which actions that lead to the precipitation of oxides
contained them. This latter observation is clearly must produce local composition gradients of the
of interest because titanium is present in participating elements between the oxidized and
case-hardening grades of steel in amounts of less unoxidized layers. This is because the atoms of
than 0.1%, and molybdenum is one of those al- the elements involved feed down the gradient in
loying elements regarded as being unlikely to be an attempt to compensate for those that have
involved in the oxidation process. Perhaps the been utilized to form the oxides. Such an effect,
atomic sizes of titanium and molybdenum com- involving both chromium and manganese, has
pared with that of iron (Ti, +36%; Mo, +10%) been reported by a number of researchers (Fig.
makes their inward or outward diffusion rates 1.9) (Ref 17, 18, 21, 22). However, it is not al-
sluggish. The same authors observed that with ways the case that chromium and manganese
samples of the same steel, carburized in the same composition gradients are present together. In
heat, one sample produced the chromium oxide, Ref 17, Gunnerson determined the composition
whereas the other sample produced manganese of the oxides at the surface of a carburized
oxide. The reason for this difference in behavior 15CrNi6 steel and came up with three different
was not established, although it was considered results (Table 1.1). This suggests that manganese
that alloy segregation differences or the presence and chromium can jointly or independently form
of varying amounts of carbide might have been oxides; much depends upon their respective
responsible. Agreeing with KaIner and Yurasov, quantities in the steel as a whole, or locally due
~un Yitang also identified the oxide as FeDerZ03 to microsegregation. If only the oxide of one ele-
10 a 20CrMnTi steel (Ref 4). The oxide Si Oat ment forms, it is reasonable to assume that the
grain boundaries and (Crx Mny)nO as pC:Cipi- matrix of the steel will be locally depleted of
tates were detected in a carburized Mn-Cr-B only that element. Arkhipov, using a 3Y2%
steel (Ref 18), and again it was found that the sil- Ni-Cr-W steel, determined that whereas the
chromium migrated toward the surface, the
nickel and tungsten migrated into the body, and
Surface it was observed that in a discrete zone, the nickel
Si <Il
content exceeded 5% (Ref 15). Manganese ap-
oxides CIl
peared to diffuse only a short distance to the

~ <Il §
E "a grain-boundaries. Robinson observed negative
'-- s '0
Cr z- a.l!? composition gradients involving chromium and
oxides '":>c

Mn .8
oxides c Table1.1 Electronprobe analysisof certain

o internal oxides
COII1positioO, "
Cr Mn Nt
L..-- 2 15 11.3 Undetectable

Fig. 1.8 Distribution of oxides at the surface of a
Cr-Mn alloy steel carburized at 930°C for 5 h.
Source: Ref 17
Adapted from Ref 6, 7
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
18 / Carburizing: Microstructuresand Properties

1.6 Table 1.2 Analysis of the surfacematerial in SAE

0 N~ 8620 steelcarburizedinendothermicgaswith
1.4 0 naturalgasadditions
1.2 ~~ 1.48% Ni
Compositloo, ..
o~ 0 0
Position Cr Mo Nl

",.. ~X
E 1.0 Surface 0.79 0.18 0.55

• ••
Subsurface 0.52 0.14 0.57
S! Mn 0.86% Cr
l5 0.8

o Source: Ref 23
~ 0.6
• ,,'I."':' s, 10- \
4( 0.75% Mn
":~ "

• :f'.
I molybdenum at the surface of an SAE 8620 steel

"Mn% -
6!l,' • Cr% (Table 1.2) (Ref 23), whereas Colombo et al.
,"' o Ni% - found that there were no diffusion gradients as-
o ': I sociated with the internal oxidation in a carbu-
o 20 40 60 80 100 rized SAE 94B17 steel (Ref 24).
Depth below surface, 11m
It is understood that much of an element mi-
grating to the surface is utilized in forming the
(CrxMnylnO oxide, and that the matrix material in the vicinity
I of the oxide remains, if not completely depleted
of that element, substantially below the average
Nominal~r, Mn for the steel in question.
1.2 Apart from what happens to the alloying ele-
t!. - -- --- - - -- - -- -7 -7 c,.. - --- ments of a steel during the oxidation process, it
E 1.0
S! Cr j Mn
may be found that the carbon content is also af-
l5 0.8 fected. In Ref 1, Kozlovskii et al. related that a

<~ 0.6 1/ / sample of a carburized 25KhGT steel exhibited a

low carbon content in the layer of internal oxida-
0.4 / I tion (Fig. UOa). Other samples of different steel
0.2 1/ compositions used in the same investigation did
not show the same decarburization effect, which
o / makes it tempting to believe that this was a case
o 10 20 30 40 50 60 70 80 of normal decarburization. However, using an
Depth below surface, 11m
SAE 94B 17 steel, Colombo et al. determined the
carbon content within the layer of internal oxida-
tion (i.e., within the outermost 20 urn) to be
1.0 0.53% (Fig. 1.10b) (Ref 24). Shcherbedinskii
Cr and Shumakov also suggested that there was a
'it 0.8 reduced carbon content in the oxidized layer
E 0-" 0.- 0 •• 01. ••• ------ o,
(Fig. 1.11), but only where the oxides have
s 0"

8 0.6
formed (Ref 5). These researchers considered
~ 0 ~n that internal oxidation took place within the car-
4( 0.4 bides, or at the carbide-austenite interfaces,
/1 which implies that a high carbon potential is a
necessary requirement of the oxidation process.
If 20 40 60 80 100 120
Other researchers have shown that a low carbon
potential (high-carbon dioxide atmosphere) most
Depth below surface, 11m favors internal oxidation. It also favors
decarburization (see Chapter 2).

Fig. 1.9 Composition gradients associated with internal

oxidation. (a) Electron probe microanalysis of manganese,
chromium, and nickel within the surface zone of 15C4rNi6 Effect on Local Microstructure
steel. Source: Ref 17. (blChromium and manganese concen-
tration gradients beneaththe internally oxidized surface of a
20MnCrB5 steel. Source: Ref 18 and 22. (e) Chromium and As yet, only the formation of the actual oxides
manganese profiles measured bymicroprobe analysis ofsteel has been discussed. However, a consequence of
SIS 2515.Source: Ref 21 internal oxidation and the composition gradients
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 19




/ - r-, .......

tf. 0.6 I tf. 0.6

~ s<:
8 8

~CIl<: 0.4 j 0.4 SAE94B17 -

o 0.2%Mo 0.28%Si
A 0.4%Mo 0.54%Ni
0.2 o 0.5% Cr, 0.5% Ni, 0.2% Mo - 0.2 0.38% Cr -
• 1.0% Cr, 0.2% Mo 0.11% Mo
.25KhGT-1%Cr,1%Mn,O.1%TI 0.05% AI
• 30 KhGT -1% Cr, 1% Mn, 0.1% TI 0.002%B

o I I I o I I
o 0.025 0.050 0.075 0.1 o 0.05 0.1 0.15 0.2 0.25
Distance from surface, mm Distance from surface, mm
w ~
Fig. 1.10 Examples of low-carbon surfaces associated with internal oxidation. (a) Source: Ref 1. (b)
Source: Ref 24

that develop during oxide formation is that the who found that the HTTP associated with the in-
material adjacent to the oxides will have its ternal oxidation in a carburized SAE 8620 steel
transformation behavior modified. Thus, instead consisted of both pearlite and bainite. Neither of
of the expected martensite, high-temperature these, only martensite, was observed in the sur-
transformation products (HTTP) can develop face of a carburized SAE 4615 steel (Ref 20).
(Fig. 1.12). The nonmartensitic microstructures, Table 1.3 indicates the extent of alloy depletion
which occupy the same area affected by internal within the matrix of the internally oxidized lay-
oxidation, are variously described as pearlite or ers of these two steels.
quenching pearlite, or either or both lower and The hardenability effect is illustrated for a
upper bainites, or mixtures of all of them. It is 17CrNiM06 steel in Fig. 1.13(a). This steel has a
likely, however, that the hardenability of the good case hardenability and is recommended for
layer and the cooling rate are each significant. A use in "driving pinions and high stressed cog
lean-alloy steel or heavy section will tend toward
a surface containing pearlite, whereas a more al-
loyed steel or lighter section will tend toward a
bainitic microstructure being formed on quench-
ing. The situation is to some extent confused by
the presence of oxides that offer substrates on
which new phases can nucleate, and by any local
stresses that develop during the quench. Which-
ever nonmartensitic microstructure is formed, it
will be comprised of ferrite and carbides, and the
rate of cooling will dictate how the carbides pre-
cipitate. There is a chance that no HTTP will
Distance from surface ~
form when the cooling rate is high or when there
Fi~. 1.11 Example of the relative distributions of
are sufficient amounts of nickel and molybde- carbon and oxygen at the surface of a carburized,
num in the matrix adjacent to the oxides. An ex- highly alloyed steel that contains a high density of
ample of this was provided by Dowling et al. carbide phase in the outer case. Source: Ref5
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
20 I Carburizing: Microstructures and Properties

wheels" (Ref 25). The illustration indicates that nal amount as a result of internal oxidation. This
even with a section equivalent to a 50 mm diam- excludes the likelihood of the internal oxidation
eter bar, some bainite will form if the manganese providing favorable sites for the nucleation of
and chromium are reduced to half of their origi- HTIP.

'. "",-....
~ ,"





Fi~. 1.12 Etching to reveal the presence of high-temperature transformation products associated
with internal oxidation. (a) Unetched. 500x (b) lightly etched in 2% nita!' 500x (c) Medium etched in
2% nita!' 500x
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 21

If the carbon content is reduced to 0.5%, for mium depletion takes place. This suggests that
example, as it was in the case-hardened sample with carbon contents over -0.15%, free ferrite is
examined by Colombo (Ref 24), then for the unlikely to be produced in sections equivalent to
17CrNiM06 steel in question, the largest section -400 mm (16 in.), but low-carbon bainite will
to avoid bainite formation in the low-carbon sur- form.
face layer is 100 mm (4 in.) (Fig. 1.13b). If the Many case-hardening steels have harden-
low-surface carbon content is accompanied by a abilities less than that of the steel used for com-
50% reduction of both manganese and chro- piling Fig. 1.13, and therefore will have more of
mium, the limiting section will be approximately a tendency to form HTTP adjacent to any inter-
37 mm (I~ in.). If, however, the manganese and nal oxidation formed at the surface, including the
chromium are completely removed to form ox- formation of ferrite . Figure 1.14 depicts the con-
ides, even light sections will likely have bainite dition for a carburized lean-alloy SAE 8620 steel
associated with the internal oxidation. In this in- with manganese and chromium depletion and
stance, the pearlite nose will be in excess of 1000 degrees of surface decarburization. By compar-
seconds. ing this with Fig. 1.13(c) it can be seen that the
Figure 1.13(c) considers the situation where
both manganese and chromium are removed Table 1.3 Semiquantitative analysis of elements
from solid solution by oxide formation, and how in material adjacent to oxides
this affects the ruling section when the carbon is Composition, 'I
also reduced. The indication here is that below Sled Cr Mn Si
-0.25% C, some ferrite will be produced at the 8620(8)
surface in all except the very lightest sections. Casemartensite 0.67 1.04 0.3
Note that with normal transformation behavior, Bainite (at oxide ) 0.37 0.6 0.2
Pearlite (at oxide ) 0.24 0.26 0.3
this steel would be unlikely to form pearlite at
the cooling rates being considered.
Case martensite 0.2 0.51 0.4
To complete the set, Fig. 1.13(d) illustrates the Martensite (at oxide) 0.1 0.22 0.3
effect that different levels of surface decarbur-
ization will have on the steel transformation (a) 0.92% Mn. 0.5% Cr. 0.11% si, 0.38% Ni. 0.16% Mo. (b) 0.52%
Mn. 0.12% Cr, 0.24% Si, 1.75% Ni, 0.54% Mo . Source: Ref20
characteristics when no manganese and chro-

Fig. 1.12 (continued) (c) M edium etc hed in 2% nital. SOOx
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
22 / Carburizing: Microstructures and Properties

8OOr-----y-----,---,------. 8oo.....----..----,----""T"""-----,

!;>600 !;> 600 I------l----+----:::::;;;;;;--+....:...::=----\j

400 t----l-------1HP\l-~-r- +_--j

1 400
{!!. 200 {!!. 200 1-------.:;:.:..::~.:::.::-=:.:.jI--'~~~--+--____1
1.5 Cr, 0.6 Mn

oL-_ _----l -L ....L. _ _----J

10 10-1 10
Time,s Time,s
(8) (b)

800r-----y-----,---,------. 800 .....----..----,----""T"""-----,


!;> !;> 6OOI-I-~:j:~:::.;~~dj

~ ~
~ ~ 400 1-----+
~ II Positionof
the peart~e a
~200t___---l__--___I---_+----'-":::::..='___l nosetime
~ 2001----1----+----+0

~0...._ - : 1 - - - l - - - - - - J - - - - - l . - - - - J
oL-_ _----l -L ....L. _ _----J

10 1()2 10-1 10
Time,s Time,s
(c) (d)
Fig. 1.13 Effect of composition variations on the transformation behavior of a case-hardening steel 17CrNiMo6. (al
Variations in Mn and Cr contents with composition 0.7 C, 1.5 Ni, 0.3 Mo. (b) Variations in Mn and Cr contents with compo-
sition 0.5 C, 1.5 Ni, 0.3 Mo. (c) Variations inC content with composition 1.5 Ni, 0.3 Mo, OCr, 0 Mn. (dlVariations inC con-
tent with composition 1.5 Ni, 0.3 Mo, 1.5 Cr, 0.6 Mn

leaner grades of steel are the more likely to form num reduced the amount of HTTP (Table 1.4)
pearlite at an oxidized surface. It is evident that (Ref 26).
the HTTP (ferrite, bainite, or possibly pearlite) • Section size and the quench severity, both of
associated with the internal oxidation depends which influence the cooling rate. The faster
upon: the cooling rate is, the better the chance of
• Steel composition and the quantities of ele- suppressing the formation of HTTP.
ments remaining in solid solution following • Increase in carburized case depth, possibly.
oxidation, including carbon. It was confirmed This is the case if an increase in case depth
by the work of Arkhipov et al. that molybde- leads to a more complete depletion of alloy-
ing elements.
Alloy Depletion and the Eutectoid Carbon
0.5%C Content. The reduction of hardenability due to
manganese and chromium depletion may not be
the only consideration. The removal of manga-
nese and chromium, or either one on its own,
~ 400 F"'----+--+--+-\--4-f-=------j would be expected to raise the eutectoid carbon
~ content in the alloy-depleted layer shown in Fig.
E F,lerrite
{!!. 200 P, peart~e
1.15, which means that the solubility for carbon
B, bainite will locally increase. Therefore, if the carbon
M, martensite
content of the surface is at C I, and it is sufficient
o'--;-__--'- ...L.-_ _--.l..,....-_--.J

10- 1 10 1()2 103

to produce free carbides in the absence of any al-
Time, S loy depletion, then where alloy depletion has oc-
Fig. 1.14 Effect of alloy depletion and carbon content curred and the eutectoid carbon content has been
on the continuous cooling transformation behavior of an raised, there will be more carbon in solution,
SAE 8620 steel with composition 0.5 Ni, 0.2 Mo, 0 Cr,0 Mn and, hence, less free carbide. Figure 1.16 depicts
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 23

Table 1.4 Effect ofdepth of nonmartensitic layeron hardness, residual stress,and bendingfatigue
strengthincase-hardened and tempered4 mmmodulus gears
Depth ofnon- MJerohardne••
Casedepth, Macrohardness, HRC martensidc intooth roots, Residual stressat adepth orl0 flIR Bending fatigue limit
Steel Process(a) mm Surfa"" Core Layer,URl HVO.2 kg/mm1 MPa kgLmmz MPa
25KhGM CN 1.0 59/61 39/44 6 560m O 77 753
20KhNM C 1.2 57/60 32136 7 5901760 0-5 0-50 71 695
25M0 5Kh05 CN 0.7 58/60 24128 3 525nOO 0 0 68 665
C 1.2 60162 22126 6 510/575 66 646
18KhGT(l ) C 1.0 59/62 32136 16 415/440 26-44 253-432 61 598
30KhGT CN 1.1 60162 40143 15 415/510 9-12 89-119 55 540
C 1.0 60162 38144 30 375/440 14-19 139-185 55 540
18KhGT(2) C 1.2 58160 32136 17 380/500 54 532
18Kh2N4VA C 1.0/1.1 58/59 40/43 17 265/575 49 482
35Kh CN 0.7 59/62 34137 100 3201510 24-45 234-441 42 412

(a) C. carburized; CN. carbonitrided. Source: Ref26, 29

such a situation where the absence of carbides in It is not clear whether internal oxidation on its
a narrow band adjacent to the oxidized layer, own is especially deleterious, or if it simply be-
which might be attributed to decarburization, haves much in the same way as does surface
could in fact be due to a shift of the eutectoid roughness, bearing in mind that the effect of sur-
carbon, there being no loss of carbon. face roughness may be offset by the presence of
compressive residual stresses in the underlying
ste el. Having said that, it is clear that any
Influence on Material Properties nonmartensitic microstructures, that is, HTIP,
associated with the internal oxidation can have a
The chemical and microstructural effects so
far described are concerns due to their potential
influence on the properties of the carburized part.

950 /.
900 .
J leI

i\, V... .
!.. .
P 800

\ ..

~ 750
0 /
aE ... ----.~ [1-.:
..,!~ _.....
---- . _.. - .. . ..
{!!. 700

Eo. original -,
650 Em. !OOdified by f - - -
Y alloy dep letion



o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
carbon content, %
Fig. 1.16 Repr esen tation of a microstructure showing
interna l oxidation with associate d high-temperatur e
Fig. 1.15 Shiftof eutectoid ca rbon co ntent due to alloy transformat ion produ cts at the surface and spheroidized
depl etion associated w ith internal oxidation ca rbides some distance from the surface
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
24 / Carburizing: Microstructures and Properties

deleterious effect on the strength properties of a • Mixed ferrite and pearlite: decarburization
part. with intermediate carbon; 200 to 300 HV
Influence on Hardness. Specified surface hard- • Pearlite: thought to exist in the HTTP of
ness values for carburized, hardened, and some steels; 300 HV
low-temperature tempered components generally • Mixed pearlite and bainite: possibly of inter-
fall in the range 58 to 62 HRC (the actual inter- mediate carbon; 300 to 400 HV
mediate range used depends upon the size of the • Bainite: probably predominates in most
component and its application). To achieve hard- HTTP layers; 400 HV
ness values of this order, it is necessary to pro-
duce an as-quenched microstructure of high- Influence on Residual Stresses. The oxides of
carbon martensite. However, a consequence of chromium, manganese, and silicon that form in
internal oxidation is that nonmartensitic micro- the surfaces of components during their carburi -
structures are likely to be formed in a narrow zation (at temperatures typically in excess of 920
zone adjacent to the oxides, resulting in a soft DC) will likely be in compression when the tem-
"skin," even though Rockwell macrohardness perature falls to room temperature. This is be-
tests might not indicate anything other than satis- cause their volume thermal contraction over the
factory hardening. A file test, on the other hand, temperature range 920 to 20 DC will be less than
would detect its presence. that of the steel in which they reside.
When a sample of the carburized steel is sec- The HTTP associated with those oxides will
tioned and prepared for metallographic examina- be in tension. This is partly due to the presence
tion , the presence of internal oxidation and of of "compressed" oxides but also due to the vol-
HTTP is clearly determined. Microhardness tests ume mismatch between the HTTP itself and the
in the surface regions of the sample will then underlying high-carbon martensite; note that the
measure the extent of softening due to the HTTP. HTTP is probably the first material to transform,
An example is shown in Fig . 1.17, where the while the adjacent martensite is the last to trans-
structure at the surface was probably bainite, or form.
predominantly bainite. Whereas the microstruc- The residual stresses through the roots of
tures in the nonmartensitic layer may be mixed, case-hardened gear teeth have been determined
or of a low or variable carbon content, the micro- for a number of steels (Ref 28), and the trend to-
hardness will give a clue as to the type of struc- ward tensile residual stresses at the surface is
ture present in the layer. Figure 1.18 provides an
indication of hardness against carbon content for
different types of microstructures. From this the
following is derived:
800 1----+---+-+_--1--
• Ferrite: due to decarburization; C - <0.2%;
microhardness, 200 to 250 HV 7001----+---+-+_

1000.-- - - - , - - - - - , - - - - , - - - - - , >
~ 500
]i 400 I--~£.,A--l-
c:i aoo J:

~ 300
2 600
100 r--r.....-q- +-+--+--+-l--+.- l----1
. 7-0L.a- oL.g--.J
10 20 30 40
Carbon, %
Distance from surface, I'ITl
Fig. 1.17 Microhardness traverses through the inter- Fig. 1.18 Effect of carbon on the hardness of variou s
nany oxidized layer of a carburized Cr-Mn-Ti steel microstructures obs erved in plain carbon a nd lean-alloy
(30KhGT). Source: Ref 7 steels . Source: Ref27
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 25

confirmed (Fig. 1.19). Additional work on the HTTP do cause reductions of fatigue life, but the
same gears showed that the greater the tensile re- amount is sufficiently small to be regarded as in-
sidual stresses were, the lower the fatigue significant. Also, there may be a size aspect
strength tended to be (Table 1.4) (Ref 26). Naito where, for instance, 15 urn of HTTP affects a
et al. observed a similar residual stress distribu- small test piece but has a smaller effect on a
tion for carburized ns SCM415 steel with inter- larger test piece.
nal oxidation and HTTP at the surface (Ref 29). The existing test data indicate that in the
Dowling's results are of particular interest be- low-cycle region of the S-N plots, there is a gen-
cause they compare the residual stresses in two eral trend toward a small adverse effect (0 to
case-hardened surfaces: one in which pearlite 12% reduction of strength at 1()4 cycles) due to
and bainite formed due to internal oxidation and the presence ofHTTP (Ref?, 17,20,23,29,32).
one where no HTTP formed (Fig. 1.20) (Ref 20). In the high-cycle region, the fatigue life can be
Influence on Bending Fatigue. Internal oxida- reduced by as much as 45% due to the presence
tion on its own does not appear to have a great of HTTP, though 20 to 35% is more common;
influence upon the bending fatigue strength of the amount depends upon the depth, the
case-hardened parts (Ref 20, 30). If anything, it microstructural content, and hence, the hardness
may have an effect similar to a small degree of of the HTTP layer.
surface roughness. If, however, the internal oxi- The influence of case hardness on the rotating-
dation is accompanied by HTTP (bainite or bending fatigue strength of carburized steels was
pearlite), the bending fatigue strength will be sig- studied by Weigand and Tolasch (Ref 33), and
nificantly reduced. they showed that there was a progressive fall of
It was earlier believed that there was a thresh- bending fatigue strength as the case hardness fell
old value of 13 urn for the depth of HTTP below below -680 HV5 (Fig. 1.22). Their work dealt
which it had no adverse effect, and that with less with test pieces where the measured hardness
than 13 um of HTTP, surface carbon content was values represented the maximum hardness at, or
the dominant variable (Fig. 1.21) (Ref 1, 31). close to, the surface. With internally oxidized
The idea that there is a threshold might not be in- parts, only the "skin" is soft; the underlying ma-
correct, though other work has indicated that the terial is martensitic and hard. Nevertheless, the
presence of less than 13 urn of HTTP has led to a trend is the same, as Fig. 1.23 suggests. From
reduction of fatigue strength (Ref 32); 11 urn of this, a guide to the potential loss of bending fa-
HTTP obtained a 15% reduction of fatigue tigue strength can be estimated (Table 1.5). This
strength. It could be that smaller amounts of might suggest that two carburized steels with dif-

Carburized Carboni1rided

20 ~-t-----ilH-----jr-r--1


~ -40

~ -20 s
~ 40
£-40 a: 20fl\--+---tIl;;;::-t--r--j

Depth, mm

Fig. 1.19 Residual stresses at the base of the teeth in carburized

and carbonitrided gears. Source: Ref 28
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
26 / Carburizing: Microstructures and Properties


~ -100

'iii -200
'gj -300
D 8620
o 8620 E-polish
04615 E-polish
-5oo+-~ ................'"'I"":-,...............
10-3 10- 10- 10
Depth, mm
Fig. 1.20 Residual stress profiles for both 4615 and 8620 materi-
als. Source: Ref 20

ferent alloy contents can have very different fa- test pieces failed with subsurface fracture
tigue lives in the absence of HTTP, but very sim- initiation points and, therefore, were not really
ilar fatigue lives in the presence of HTTP. The relevant to the study of a surface condition; these
results of Brugger's study, for example, showed test pieces said more about the subsurface
that the fatigue limits for case-hardened properties. After replotting the data using only
20MnCr5 and 15CrNi6 were 680 and 780 MPa, those points that represented surface initiated
respectively, when there was no HITP present at failures, a distinct difference was noted in the
the surface (Ref 34, 35). When HTTP was present upper finite life part of the S-N curve, shown
in similar amounts, the fatigue limits of the two in Fig. 1.24(a); it is suggested that the fatigue
steels fell in the range 520 to 540 MPa. curve representing the specimens with HITP at
Gunnerson found no difference in high-cycle the surface might have a double knee. Naito et
fatigue limit between samples that had HTTP aI. also observed a possible double knee effect
reaching to a depth of 15 to 17 11m, and samples for the samples having HTTP at their surfaces
internally oxidized to -15 11m without HITP (Fig. 1.24b) (Ref 29). These are interesting
(achieved by inhibiting the formation of HTTP observations, the significance of which is not
with ammonia additions late in the carburizing clear, apart from the fact that both projects em-
cycle) (Ref 17). This finding is contrary to other ployed rotating beam test pieces. However, what
reported results. Unfortunately, many of Gunnerson's
'"EE ~ f
~ ..>< 800 ::E
9 00 ::E
~90 "'~.," Ok= 1

~., ~., ~
fg> 60
~ 0> 600 .J
f80 800:1!!
c: I
c: ~ ~
g> 40 400 i


--- ~
20 200
200 400 600 800 1000
6O~_ _:;;,:;-_----:::o---~---:-::!
0.6 0.7 0.8 0.9 Hardness, HV5
Surface carbon content, %
Fig. 1.22 Rotating beam fatigue strength of case-
Fig. 1.21 The effect of internal oxidation on the fa- hardened 12 mm diam specimens, notched and
tigue strength of carburized 25KhGT steel. With this steel, unnotched. The line for carburized gears shown in Fig.
internal oxidation in accompanied by a decrease in sur- 1.23 is superimposed (converted to rotating bending fa-
face carbon. Source: Ref 1 tigue). Source: Ref 33
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 27

1000 r - - - - - , - - - - - , - - - - - - - , , . . . . - - - - , 80,....----r-----,,....---r-----,
~ 900 I------'-.l-~~\r_-f____,,£... 20Kh~M (10= 7 JlITl) 80 E
S£ 25Mo ~h05(10=6 JlITl) ~ 700
-_8 18KhGT (10=16 JlITl) 701---+---+---+--I~q

+-~~~~~_:t+=-----+~-~~70 Ig

;;; I
~ ~I-_-+ _ _~~:"-+-_~600 ~
1§ 700 t----+---~~f___'\r_--+---___=l60
[ .S ~ 5
~ u.~

s~ ~O
I ~
501--A----J+---+---=1 500
~ 50 ~
~ c
ue ~

40 L--_--l._ _....L _ _L-_..::J 400

200 300 400 500 600
Minimummicrohardness of surface
of gear tooth roots, HV 0.2

w ~
Fig. 1.23 Effect of internal oxidation and surface microhardness on the fatigue properties of 4 mm modulus gears. See
also Table 1.4. 10, internal oxidation. (a) Fatigue strength plots for 4 mm modulus gears. Information on case-hardened
gears given in Table 1.4. Source: Ref 26. (b) Effect of tooth root surface hardness on the bending fatigue limit of 4 mm modu-
lus gears

the two sets of results do seem to suggest is that as not being detrimental, and even beneficial, to
up to, or at, about 1()4 load cycles there would be contact durability under slide-roll test conditions.
essentially no difference between the samples Figure 1.25 presents the contact fatigue life as a
with and without HTTP. function of the amount of material removed from
With respect to grain size, it is interesting to the surface of the test disks prior to testing. It
note the findings of Pacheco and Krauss in Ref shows that there was generally no loss of contact
30: "Fine grained gas carburized specimens tol- fatigue resistance until the internal oxidation and
erate the presence of inter-granular oxidation and nonmartensitic microstructure had been re-
have better fatigue resistance than a coarse moved, the depth of HTTP penetration being 25
grained specimen without surface oxidation." In to 37 urn, Where the soft layer had been com-
their work, which compared plasma- and pletely removed, lower values of fatigue were re-
gas-carburized test pieces, the gas-carburized corded. As a result of supporting tests, these re-
specimens contained internal oxidation to a searchers considered that plastic deformations
depth of -13 urn without the HTTP associated within the HTTP layer, which would bring about
with internal oxidation; this supports the idea a more favorable distribution of contact load,
that, on its own, internal oxidation is not particu- could only account for part of the difference
larly damaging. shown. In tests where the slip was about 30%
Influence on Contact Fatigue. Most case- and the contact load was 2390 MPa (347 ksi),
hardened gears are precision ground before go- they found that the unground surface had a lower
ing into service; therefore, the effect of internal coefficient of friction than the ground surface did
oxidation on contact fatigue durability need not be
considered for them. There are, however, a few Table 1.5 Loss of fatigue strength
types of gears and a number of other components Loss of fatigue streogth,
that enter service in the unground, or perhaps in Skin hardness, HVO.2 Fatiguelimit, kglmmz 'I

the lightly lapped, condition for which the influ- 700(a) 80(b) o
600 72.5 9
ence of internal oxidation is pertinent. Unfortu- 500 65 19
nately, there are little data available on the sub- 400 -58 28
ject, and these tend to conflict. Adverse effects 300 51 36
250 48 40
have been reported in Ref 3, whereas in Ref 36
Sheehan and Howes, working with case- (a) 700HVis an extrapolation;seeFig. 1.23.(b) 80 kglmrn2 is taken to
hardened SAE 8620, regarded internal oxidation represent 100%fatiguestrength.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
28 / Carburizing: Microstructures and Properties

(0.068 and 0.080 respectively) ; they reasoned, seem to yield results that are significantly influ-
however, that this difference was not large enced by the presence of internal oxidation and
enough to account for the differences in fatigue any accompanying HTIP.
strength that were observed. It was concluded, Kozlovskii et aI. stated that there was no nega-
although the test did not really confinn it, that it tive effect of on the impact strength of case-
was likely that an "as-heat-treated" surface hardened steels due to HTIP and indicated that
would develop a more stable oil film than either there was no clear separation of bend test results
a ground or electropolished surface. whether the HTIP was more or less than 13 urn
Influence on Bending and Impact Fracture (Ref 1). Diesburg et aI. concluded that fracture
Strength. The fracture testing of case-hardened was independent of the structure state in the
samples by slow bending or by impact (plane, outer 0.25 mm of the case and was more depend-
notched, or precracked spe cimens) does not ent on the residual stresses and microstructure
away from the surface (Ref 37). Tests showed
that fracture toughness increased with distance
from the surface (Ref 38), and that crack initia-
120 tion was influenced by case hardenability (the
ability to produce martensite throughout the
'"E 110
1100 case); crack propagation, however, was more
.>t; ~ influenced by the core properties and composi-
::::!i tion (Ref 39). Feu agreed that composition was
1000 gf
Ii 100 I----Ti---''.-"''''---t-----l e important with respect to bending and impact,
~ 1ij
the resistance to these being enhanced by nickel
'g ooo'g
90 I----t------;.,-'r-.---"vt--------; 1l

\ "
••. .'1.... 800 ~
80 • 8amples with HTTP -+-~?----l It
It ° samples wiIhoul HTTP Depth 01 surface material removed. I'm
- Surfaee-initialed failure 1_...: 1_..;;.
••• - Subsurface failure
70 '-----'--------.l------J700 40'
b ---'-
10.:.. 2:.:;
!_..;;.50.:.. 100.:.:.,.I_ _!o

10" 10S 10S 107 10S R • reheatedand

Cyclesto failure ("-'!) 30 quenched
D • direct quenched
20 ~----':~

~ 1200
~ 1000 I---t--".;i'\<-----..-----"'-=F-'-""""',..."I--
~ 800 I-----l---+---,---t-£--~---:~
J 6001-----+----+---+----+-----<>-
F, fish eye origin

10" lOS 106 107

Cyclesto failure ("-'!)
(b) 2
Fig. 1.24 Effectof internal oxidation and high-temper-
ature transformation products on the high- and low-cycle
bending fatigue strength. (a) Fatigue data on rotating
beam tests, 6 mm outside diameter test section, quench 1L., 1 - - -.1...---.01
1o-"""""" 10
860 DC into oil at 200 DC. Steel composition: 0.75 Mn,
0.86 Cr, 1.48 Ni, 0.04 Mo. Samples w ith HTIP: 7 failed at Depth 01surface material removed. 0.001 in.
surface, 4 ran out , 16 failed subsurface. Samples without
HTIP: 10 failed at surface, 2 ran out , 23 failed subsurface. Fig. 1.25 Contact fatigue life in relation to the amount
Replotted from Ref 17. (b) Fatigue data for samples with of material removed from carburized SAE 8620 samples
and without internal oxidation, quenched from 850 DC prior to testing . The hatched band represents the depth of
into oil at 70°C, temper at 180°C. Steel composition: internal oxidation beneath the original surface of speci -
0.74 Mn, 1.01 o.
0.18 Mo. Source: Ref 29 mens . Source : Ref 36
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation /29
Internal Oxidation / 29

and chromium
chromium (Ref (Ref 40).
40). Others,
Others, however,
however, con- con- Terms
Terms for for adhesive
adhesive wear wear include
include scoring,
sidered that
that static
static bending
bending and and impact
impact strength
strength scuffing, galling,
scuffing, galling, and
and seizure.
seizure. Fretting
Fretting is
is aa form
were related
related toto the
the case
case depth
depth (Ref
(Ref 41).
41). OnoOno et et of adhesive
of adhesive wearwear forfor which
which the
the relative
relative move-
al., using
using unnotched
unnotched impact impact and and bend
bend tests,
tests, ment is
ment is minute,
minute, as as in
in the
the vibration
vibration between
between aa key
showed that that 930°C
930 °C direct-quenched,
direct-quenched, gas- gas- and keyway,
and keyway, for for example.
example. Abrasive
Abrasive wear,
wear, oror nor-
carburized test pieces
carburized pieces with 15 15 um
\ua ofof HTIP
HTTP per- per- involves the removal of
mal wear, involves of particles from
formed better than did
formed did 1040
1040 EC vacuum- vacuum- surfaces by asperity
the mating surfaces asperity shearing
shearing due
due toto
carburized test pieces
carburized pieces with no no internal
internal oxidation
oxidation asperity collisions
asperity collisions oror collisions
collisions with loose
loose debris
or HTTP (Ref(Ref 32).
32). AA 930
930 EC EC reheat-quenched,
reheat-quenched, passing between the sliding sliding surfaces.
surfaces. Wear pr0-pro-
vacuum-carburized set
vacuum-carburized set ofof test
test pieces
pieces produced
produced cesses are
cesses are appreciably
appreciably influenced
influenced by anyany lubri-
even better results (Fig. 1.26). Unfortunately,
even better results (Fig. 1.26). Unfortunately, no no condition.
cant present, its quality, and its condition.
data were provided for any
data were provided for any reheat-quenched,reheat-quenched, When internal oxidation is present on its own
gas-carburized samples,
gas-carburized samples, so so aa real
real comparison
comparison at a case-hardened surface, one would expect it
could not
could not be
be made.
made. Nevertheless,
Nevertheless, the the results
results im- im- to inhibit the adhesive wear due to
to to its
plied that factors other than internal
plied that factors other than internal oxidation oxidation intermetallic nature.
nature. Also, as Sheehan and and
and the
and the presence
presence of of HTIP
HTTP were were more
more significant
significant Howes suggested in Ref 36, 36, the oxide could as-
with respect to bending and impact fracture
with respect to bending and impact fracture tests. tests. sist in the lubrication process; as it is, the oxide
Influence on Wear Resistance. The
Influence on Wear Resistance. The deteriora- deteriora- coefficient of friction. The oxide,
will lower the coefficient
tion of
tion of aa surface
surface by by wear
wear can can be
be due
due to to either
either oror however, might favor abrasive wear and eventu-
both of
both of two
two processes:
processes: adhesive
adhesive wearwear and and abra-
abra- ally crumble under the action of sliding, thereby
sive wear.
sive wear. Adhesive
Adhesive wear,
wear, as as the
the name
name suggests,
suggests, producing loose particles (debris). In such an in-
is adhesion that occurs when the
is adhesion that occurs when the pressure and pressure and stance, the depth of wear will likely be only the
heat generated during sliding cause
heat generated during sliding cause small areas small areas depth
depth ofof internal
internal oxidation:
oxidation: 10 10 to
to 20
20 urn.
of one
of one ofof the
the surfaces
surfaces to to chemically
chemically bond bond to to the
the When the internal oxidation is accompanied
other surface. The relative motion
other surface. The relative motion breaks the breaks the by the HTTP, the situation in terms of wear- wear-
bond, but
bond, but not
not necessarily
necessarily (perhaps
(perhaps rarely)
rarely) at at the
the resistance changes. Both adhesive and abrasive
original junction; therefore, metal is
original junction; therefore, metal is transferred transferred wear resistances are related to the surface surface
from one
from one surface
surface to to the other. What
the other. What happens
happens to to hardness (Fig. 1.27, 1.28). As the hardness
the transferred particle then depends upon aa
transferred particle then depends upon falls, the wear rates will increase, considering
number of of factors.
factors. ItIt may
may remain adhered to
remain adhered the
to the that HTTP will work harden, and martensite
other surface, or it may separate to become loose will soften a little as pressure and temperature
other surface, or it may separate to become loose
debris. conditions approach those that cause adhesive adhesive

Total energy
(initiation plus 2000
2000 4
propagation) Ultimate bending
-, Crack initiation
£l!! 1500
energy c
c: o
/ '6

, ,e-............
c: S
o "8
~ 1000
1000 ' a
I,§5~ ....... e'" "-"-
10 500
Crack propagation
oL-.+_---\!::--_+-_-!:_ _..J _ _...J.._ _--I,,_ _.....!,---I

A a U D3 B A C D3
Load Load
Fig. 1.26 Variations of (a) unnotched impact energy and (b) bending strength (15 x 60 x 2 mm). A: vacuum-carburized,
1040 °C, reheat quenched. B: vacuum-carburized, 1040 °C, direct quenched. C: vacuum carburized, 930 °C, reheat
quenched. D3: gas-carburized, 920 °C, direct quenched. Source: Ref 32
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
30 / Carburizing: Microstructures and Properties

wear. In case-hardened parts, high-carbon Measures to Eliminate or

martensite at over 700 HV offers the best resis- Reduce Internal Oxidation
tance to wear. Other phases, such as proeutectoid
ferrite or austenite, can be tolerated in the mar-
tensite without adverse effects, provided those It is more or less impossible to control the en-
phases are indeed in small quantities, are fine and dothermic carburizing atmosphere to eliminate
well distributed, and don't significantly reduce the the formation of internal oxidation at the sur-
hardness (Ref 42). If the lITfP is bainite at -400 faces of case-hardened parts. The use of alterna-
HV or ferrite at, for instance, 250 BY, then the tive oxygen-free atmospheres or of vacuum-
wear resistance of the case-hardened layer will be carburizing or plasma-carburizing processes
seriously affected. Whether or not the wear pro- might not be economically viable or as flexible
cessesstop or slowdownonce the lITfP has been and convenient as the conventional process. Con-
removed by wear depends on the actual system in sequently, the endothermic generator will con-
which the damage is taking place. Wear processes tinue to provide the carrier gas for the carburiz-
can virtually stop. ing process for many years to come . The
problem of internal oxidation will, therefore,
persist. What can be done about it? The ap-
proaches can involve steel design, process con-
trol, mechanical or chemical removal, or design
E 500 of components that acknowledge its presence.
~ ~Jf ~ Steel Design. The formation of internal oxida-
~ 400
,/ - tion is related to the presence of certain alloying

:ll Vo V elements and their quantities in solid solution
~ °OJ within the steel. The problem is that silicon is

~ 200
~ V used to kill case-hardening grades of steel, and
the nominal0.25% siliconcontent by far exceeds
the maximum required to produce internal oxi-
8 00 .4 C steel dation. It also lowers the maxima for the other
E 100 - elements, such as manganese and chromium.
:::l .0.4 C, 0.1 Cr
.~ I I The replacement, or partial replacement, of oxide-
~ 00
I forming elements with elements that have atomic
200 300 400 500 600 700 800
Hardness, HV numbers greaterthan that of iron could contribute
(8) to reducing the effect. Such elements are nickel
and molybdenum, and maybetungsten.
/ P- The basic alloy, therefore, would contain:
oi ..... • Carbon: to adjust the core strength, though
~ 14 /0 01 not enough to adversely affect the develop-

"" 12 •
§ 10

/ ••
/ ,I
0. 0 2 0 r - - - - - - - - - - - - - - - .
'CD / / oCarburized - 0.018 x White cast iron
• Nodular iron
0 •V • Carbonitrided _ • • • Steelbefore wear
7: 0.016

• • I I o
• Steelafter wear
200 400
Surface microhardness, HV(100 9 load)
600 800 1000
§ 0.010
~ 0.008
Fig. 1.27 Relationship between surface hardness and
seizure. (a)Relationof hardness, HV, with maximumcon- 0.006
tact stress, am,x, when destructive seizure occurs for O·OO4r-."k;;.---.....-k..------,;*"-"*"-~~_...,~
200 400 600 800
through hardened or induction hardened steels. Source:
Ref42. (b)Variationof seizing load with microhardness of
° Hardness . HV

the ?uter layer of carburized and carbonitrided samples. Fig. 1.28 Effect of hardness on wear. Note that wear
Radiusof curvature, 2.5 cm; slip speed, 1.05 m/s; rolling resistance of the steel sample only fitsthe general pattern
speed 2.28 mls; specific slip, 1.7; gage width of roller, 0.5 if the hardness of the work-hardened surface is consid-
em. Source: Ref43 ered. Ref44
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation /31

ment of favorable residual stresses or to en- that as the carbon potential rises, the oxygen
courage unwanted distortion or growth content of the atmosphere decreases. Contrary to
• Nickel: to contribute primarily to the tough- this, KaIner and Yurasov found that with the
ness of both the case and the core lean-alloy steels they studied, 0.5% molybdenum
• Molybdenum: to provide case and core did not prevent or reduce the internal oxidation
hardenabilities, and also to suppress any of chromium or manganese steels (Ref 3).
HTIP should internal oxidation occur Chatterjee-Fischer tends to believe that as long
as the accepted norm for silicon content is pres-
Added to these, optimum amounts of deoxidizing ent in the steel, internal oxidation will occur irre-
and grain-refining elements must be included, spective of how the other elements are adjusted
bearing in mind that these may be elements with (Ref 6). Nevertheless, the presence of any alloy-
atomic numbers lower than that of iron. ing element must be beneficial if, for the section
Hardenability can be assisted further with moder- sizes being considered, it can inhibit the forma-
ate additionsof manganese and chromium, but less tion of HTIP.
than 0.5% of each should be added when silicon is Process Control. Prior to the heat-treatment of
present in amounts typically 0.25%. The aim is to parts in an endothermic atmosphere, precautions
accept that internal oxidation will occur to some should be taken to ensure that the surfaces are
extent but avoid the formation of significant com- free from metal oxides (scale or corrosion prod-
position gradients. ucts) and certain lubricants. Such surface impuri-
According to Kozlovskii et al. in Ref I, nickel ties can contaminate the furnace atmosphere;
at 1% appears to be incapable of providing the they appear to influence the amount of oxide
hardenability necessary to inhibit bainite forma- subsequently formed and the depth to which it
tion at the boundary regions when internal oxida- penetrates (Ref 3). Furthermore, the quality of
tion and diffusion gradient have occurred. The the machined surface can have a bearing on how
effect of nickel on the hardenability of a steel well a surface responds to carburizing. Machining
picks up somewhat when its content exceeds with sharp tools, along with the use of good qual-
-1.5%, especially in the presence of molybde- ity cutting lubricants, makes for uniform machin-
num. When nickel content is higher and direct ing and uniform carburizing (Ref 45).
quenching techniques are used, unacceptable Although it is clear that internal oxidation
quantities of austenite can be retained in the takes place during the carburizing operation, it
outer part of the case. With reheat quenching, on can nevertheless be intensified during any
the other hand, higher nickel content values can high-temperature tempering cycle (carried out in
be tolerated, although a downward adjustment of air to facilitate intermediate machining opera-
the surface carbon content might be prudent. tions) (Ref 15); it can also be intensified during
Internal oxidation has been reported in reheating operations, normally under atmo-
chromium-molybdenum steels, but in such cases sphere, prior to quenching. Heating in air, espe-
the molybdenum content has generally been low cially to the quenching temperature, will cause
(-0.2%). The introduction of 0.5% (or more) scaling to occur, which might be found to scale
molybdenum is claimed to be very beneficial, es- off the internally oxidized layer. Subsequent shot
pecially in those steels where molybdenum has blasting would then remove both types of oxide.
been sensibly balanced with elements having a This approach to internal oxide removal, while
positive oxygen affinity that also impart neces- possible, is not really recommended due to the
sary hardenability. For example, internal oxida- difficulty of precise process control and the risk
tion to a depth of 14 to 20 urn was observed in of inducing decarburization if the reheating pe-
steels with molybdenum to chromium ratios of riod is lengthy.
up to 0.4, but not in steel with a molybdenum to As stated previously, internal oxidation on its
chromium ratio of 1 (Ref 1). With a nickel- own might not be particularly damaging to fa-
molybdenum steel where the molybdenum to tigue resistance, whereas the presence of the
chromium ratio was 0.4 and the nickel content HTTP can be harmful. The aim, therefore,
was over 1%, element impoverishment (indi- should be to suppress the formation of HTIP. By
cated by dark etching constituents in the increasing the cooling rate during the quenching
metallographic sample) was only observed with operation, it may be possible to achieve this aim
a reduced carbon concentration in the case. and thereby have only martensite associated with
Therefore, it would seem that high-carbon sur- the oxides. Increasing the cooling rate during
faces are less prone to oxidation, bearing in mind quenching could lead to distortion or growth
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
32 / Carburizing: Microstructures and Properties

problems, however. The critical cooling rates who reduced or eliminated near-surface bainite
have been established for a number of steels and by increasing the carbon potential to 0.9 to 1.2%
the optimum quenching parameters for those during the last 20 minutes of carburizing (Ref
steels were determined (Ref 46). Unfortunately, 48). The method, it seems, does not form new
as the size of the component increases, it be- carbides but raises the carbon content of the solid
comes more difficult to achieve the cooling rates solution and, therefore, the hardenability of the
necessary to ensure the suppression of HTfP. outer surface.
For example, Fig. 1.29 indicates the cooling time Oxide Removal. The knowledge that internal
to 400 °C for the fillets of gear teeth of different oxidation will occur, and its confirmation by
sizes (Ref 27). When these are used in conjunc- means of suitable test pieces that have been car-
tion with the continuous cooling transformation burized and hardened along with the components
(CCT) diagrams of Fig. 1.13 and 1.14, one can they represent, enables a manufacturerl
loosely assess how component size might restrict heat -treater to develop an acceptable method for
the option to increase cooling rate to suppress the removing the oxide and HTfP that have formed.
formation of HTfP. Such methods as electropolishing, electrochemi-
Surface hardenability and strength can be re- cal machining, honing, grinding, grit blasting,
stored by the introduction of ammonia into the shot blasting, or peening might be considered.
carburizing chamber for a short period at the end Which method is used will depend upon what is
of the carburizing cycle. Gunnerson suggested a available and what the negative effects might be;
5 to 10% ammonia addition for ten minutes be- for example, grinding to remove oxides can in-
fore the end of carburizing (Ref 17). Using the duce tensile residual stresses of a magnitude in
same guidelines, Kozlovskii et al. determined excess of the tensile stresses associated with in-
that a nitrogen content of 0.1 to 0.2% was ternal oxidation.
achieved in the outer 0.05 mm of surface (Ref 1). If it is essential to have an oxide-free surface,
That was effective in removing the tendency to but its removal is unacceptable for some reason,
form HTfP in all but the leanest steels, and for then it will be necessary to consider the use of al-
the leanest steels only when they were fine ternative carburizing processes, for example,
grained. Gu et al. determined that a nitrogen con- vacuum or plasma carburizing. Nitrogen-base at-
tent of 0.1% in the outer 1 mm was the optimum mospheres and exothermic-based atmospheres
amount (Ref 47); when nitrogen exceeded 0.1 %, would be expected to reduce the amount of oxi-
the crack-growth rate property increased. An al- dation but would not eliminate it.
ternative action was used by Zinchenko et al., Part Design. It is difficult to say how the results
of tests involving samples with test sections of 6
to 8 mm and case depths of 0.5 to 1.5 mm trans-
r\ Rimor body Computedtime,s,at given diametricalpitch:
late to real-life components. However, these cur-
r' '" thiclmoss,in. I 3 5 rent assumptions are not unreasonable: conven-
I I 1 106 68 59 tional gas carburizing produces internal
I I 6 225 133 108 oxidation, Internal oxidation will likely be ac-
U 12 225 110
-=------===----===----=-=-=---- 135
companied by HTfP, and fatigue strength will be
Rimor body Computedtime, S, at givendiametricalpitcb:
approximately 25% lower than for an oxide-free
thickness,IlL I 3 5 counterpart. Designers may cope with these out-
, I 1 155 79 65 comes, as gear designers have done for many
I I 6 419 161 120
L-.: _12
years, by having conservative basic allowable
440 163 120 _
design stresses, or by accounting for the poten-
tial of this surface condition in their calculations.
r\ Rim or body Computedtime, S, at givendiametricalpitch:
rl ~ thickness in. I 3 5 (Internal oxidation is acknowledged in gear stan-
, I 1 155 93 81 dards like AGMA 2001, C95, or ISO 6336-5.2
I • I 6 880 758 733 [1995], where acceptable amounts in relation to
U 12
-------------- 2400 2290 2180
tooth size are given. These amounts probably re-
Fig. 1.29 An indication of cooling times for gear fer to the depth of the HTfP.) One should con-
shapes. (a) Time for gear tooth fillet surface to cool from sider that processes, such as vacuum carburizing
800 to 400 °C during oil quenching. (b) Time for gear and plasma carburizing, while capable of pro-
tooth center on the root circle diameter to cool from 800 ducing carburized surfaces that are free from in-
to 400°C during oil quenching. (e) Time for rim or body
center to cool from 800 to 400°C during oil quenching. ternal oxidation, can produce surfaces in which
Source: Ref 27 the austenitic grain size is coarse and the quan-
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 33

tity of retained austenite is excessive (Ref 30). • Preprocess considerations: Consider a steel
The designer may have to allow for these also. grade with sufficient matrix alloy in the de-
pleted layer to give martensite on quenching.
• In-process considerations: A late increase of
carbon potential, or a late addition of ammo-
Summary nia to the furnace chamber can offset the ef-
fect. Employ a faster quench to suppress
Internal Oxidation HTTP formation, but watch for distortion.
Internal oxidation is a surface effect due to the • Postprocess corrections: Grinding, grit blast-
oxidation of certain elements in the steel (mainly ing and/or shot peening can be used to re-
Mn, Cr, and Si) during carburizing. It is unavoid- move HTTP.
able in conventional gas carburizing, but can be • Effect on properties: HTTP is soft and has an
avoided in vacuum carburizing processes. adverse effect on surface residual stresses and
bending fatigue resistance-up to about 35%
• Preprocess considerations: Choice of steel reduction in extreme eases. If shallow, HTTP
grade, a small uniform grain size (normal- might have a positive effect on the contact fa-
ized), and essentially clean parts are factors. tigue of unground surfaces.
• In-process considerations: The depth of oxi- • Standards: No specification or guideline. As-
dation relates to case-depth, and the composi- sume that HTTP should not exceed the maxi-
tion of the carburizing atmosphere. mum depth of internal oxidation as judged
• Postprocess considerations: Internal oxida- metallographically. See "Internal Oxidation."
tion can be removed by grit blasting and/or
shot peening.
• Effect on properties: Internal oxidation possi- REFERENCES
bly has a minor negative effect on bending fa-
tigue, and a positive effect on contact fatigue. 1. I.S. Kozlovskii, AT. Kalinin, AY. Novikova,
See also "High-Temperature Transformation E.A Lebedeva, and AI. Feofanova, Internal
Products." Oxidation during Case-Hardening of Steels in
Endothermic Atmospheres, Met. Sci. Heat
The following are ANSI!AGMA standards: Treat.(USSR) (No.3), March 1967, p 157-161
2. L.H. Fairbank and L.G.w. Palethorpe, "Con-
Gradel Grade 3 Case depth trolled Atmospheres for the Heat Treatment of
0.0007 in. 0.0005 in. <0.030 in. Metals," Special Report 95: Heat Treatment of
0.0010 in. 0.0008 in. 0.03()...{).059 in. Metals, Iron and Steel Institute, London, 1966,
0.0015 in. 0.0008 in. 0.059~.089 in.
0.0020 in. 0.0010 in. 0.089~.l18 in. p 57-69
0.0024 in. 0.0012 in. >O.l18in. 3. V.D. Kalner and SA Yurasov, Internal Oxida-
tion during Carburising, Met. Sci. Heat Treat.
(USSR) (No.6), June 1970, p 451-454
Grade 1 has no specification. Recovery by shot 4. Y. Sun, D. Xu, andJ. Li, The Behavioroflnter-
peening is acceptable with agreement of the cus- nal Oxidation in Gas-Carburised 20CrMnTI
tomer. Note that ISO 6336-5.2 is similar to this Steel, Heat Treatment and Surface Engi-
(ANSI!AGMA 2001-C95). The equivalent grades neering, Proc. Sixth International Congress,
are ML, MQ, and ME.
Chicago, 1988
5. G.V. Shcherbedinskii and AI. Shumakov, In-
High-Temperature ternal Oxidation of Excess Carbides in Alloy
Transformation Products Steels during Carburising, Irv: Akad. Nauk
SSSR. Met., May/June 1979, (No.3), P
When associated with internal oxidation, 193-199
high-temperature transformation products 6. R. Chatterjee-Fischer, Internal Oxidation dur-
(HTTP) are a surface effect. The formation of in- ing Carburising and Heat Treatment, Metall.
ternal oxidation products locally denudes the Trans. A, Vo19, Nov 1978, P 1553-1560
matrix of certain elements. When HTTP are 7.I.Y. Arkhipov, V.A. Batyreva, and M.S.
present, they often extend deeper than the inter- Polotskii, Internal Oxidation of the Case on
nal oxidation, especially with leaner steel grades. Carburised Alloy Steels, Met. Sci. Heat Treat.
HTTP can be avoided or controlled. (USSR), Vol 14 (No.6), June 1972, p 508-512
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
34 / Carburizing: Microstructures and Properties

8. C. Van Thyne and G. Krauss, A Comparison of Atmospheres, ASM International, 1995, p

Single Tooth Bending Fatigue in Boron and 55-60
Alloy Carburising Steels, Carburising: Pro- 21. B. Hilldenwall and T. Ericsson, Residual
cessing, and Performance, G. Krauss, Ed., Stresses in the Soft Pearlite Layer of Carbu-
ASM International, 1989, p 333-340 rised Steels, 1 Heat Treat., Vol 1 (No.3), June
9. R. Chatterjee-Fischer, Zur Frage de Rand- 1980, p 3-13
oxidation bei Einsatzgeharteten Teilen, 22. Y.E. Smith and G.T. Eldis, ''New Develop-
Kritische Literaturaus Wertung, Harterei ments in Carburised Steels:' Climax Molybde-
Technische Mitteilungen, (No. 28) Nov 1973, num Co., Michigan, 1975
p259-266 23. G.H. Robinson, The Effect of Surface Condi-
10. C. Dawes and RJ. Cooksey, "Surface Treat- tion on the Fatigue Resistance of Hardened
ment of Engineering Components," Special Steel, Fatigue Durability of Carburized
Report 95: Heat Treatment of Metals, Iron and Steel, American Society for Metals, 1957,
Steel Institute, London, 1966, p 77-92 p 11-46
11. E. Mitchell, R.J. Cooksey, and C. Dawes, 24. R.L. Colombo, F. Fusani, and M. Lamberto,
"Lean-Alloy Carburising Steels:' Publication On the Soft Layer in Carburised Gears, J. Heat
114: Low-Alloy Steels, Iron and Steel Institute, Treat., Vol 3 (No.2), Dec 1983
London, 1968,31-36 25. C. W. Wegst, Stahlschliissel, Verlag
12. B. Edenhofer, Progress in the Technology and Stahlschllissel Wegst KG, Marbach, Germany,
Application of In-Situ Atmosphere Production 1977, p 14
in Hardening and Case-Hardening Furnaces, 26. I.Y. Arkhipov, M.S. Polotskii, A.Y. Novikova,
Proc. ofthe Second International Conf. in Car- .SA Yurasov, and Y.F. Nikanov, The Increase
burizlng and Nitriding with Atmospheres, in the Strength of Teeth of Carburised and
ASM International, 1995, p 37-42 Carbonitrided Gears, Met. Sci. Heat Treat.
13.ASM Metals Reference Book, American Soci- (USSR) (No. 10), Oct 1970, Plenum Pub-
ety for Metals, 1981 lishing, p 867-871
14. K. Okasaki et al., Effect of Titanium on 27. G. Parrish and G.S. Harper, Production Gas
Recrystallisation and Grain Growth of Iron, J. Carburising, Pergamon Press, 1985
Jpn. Inst. Met., Vol. 1 (No. 39), 1975, P 7-13 28. I.Y.Arkhipov and Y.A. Kanunnikova, Residual
15. I.Y. Arkhipov, Internal Oxidation of Steel Stresses in the Teeth of Quenched Gears, Met.
18Kh2N4Va,Met. Sci. Heat Treat. (USSR), Vol Sci. Heat Treat. (USSR), (No. 11), Nov 1970, P
15 (No.7), July 1973, Plenum Publishing, 909-913
622-624 29. T. Naito, H. Ueda, and M. Kikuchi, Fatigue
16. N. Murai, T. Tsumura, and M. Hasebe, "Ef- Behaviour of Carburised Steel with Internal
fect of Alloying Elements and Oxygen Po- Oxides and Non-Martensitic Microstructure
tential at the Equilibrium Carbon Content in Near Surface, Metall. Trans. A, Vol 15, July
Gas Carburising," presented at the Tenth 1984, P 1431-1436
Congress of the International Federation for 30. lL. Pacheco and G. Krauss, Microstructure
Heat Treatment of Surface Engineering, and High Bending Fatigue Strength of Carbu-
Brighton, 1996 rised Steel, 1 Heat Treat., Vol 7 (No.2), 1989,
17. S. Gunnerson, Structure Anomalies in the Sur- p77-86
face Zone of Gas-Carburised Case Hardened 31. W. Beumelburg, Der Einfluss von
Steels, Metal Treatment and Drop Forging, Vol Randoxydation auf die Umlaufbiegefes-
30 (No. 213), June 1963, p 219,229 tigkeit und Statische Biefestigkeit
18. Report No. I.S.l-888, Climax Molybdenum einsatzgehiirteter Proben, Harterei Technische
19. S. Preston, Influence of Vanadium on the Mitteilungen, Vol 25 (No.3), Oct 1970, P
Hardenability of a Carburizing Steel, Carbu- 191-194
rizing: Process and Performance, G. Krauss, 32. H. Ono, K. Okamoto, and Y. Nishiyama,
Ed., Conf. Proc. (Lakewood, CO), ASM Inter- "Some Mechanical Properties of Vacuum
national, 1989, p 191-197 Carburised Steel," Paper 32, presented at Heat
20. W.E. Dowling, Jr., W.T. Donlon, ws. Copple, Treatment 1981,Sept 1981
and c.Y. Darragh, Fatigue Behavior of Two 33. H. Weigand and G. Tolasch, Dauerfes-
Carburized Low-Alloy Steels, Proc. Second tigkeitsverhalten Einsatzgeharter Proben,
Int. Conf. on Carburizlng and Nitriding with Hiirterei Technische Mitteilungen, Vol 22 (No.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Internal Oxidation / 35

4), Dec 1967, p 330--338 (in German; also 47. C. Gu, B. Lou, X. Jung, and F. Shen, Mechani-
available in English) cal Properties of Carburised Co-Ni-Mo Steels
34. H. Brugger, Effect of Material and Heat Treat- with added Case Nitrogen, 1 Heat Treat., Vol7
ment on the Load Bearing Capacity of the Root (No.2), 1989, P 87-94
of Gear Teeth, VDI-Berichte, Vol 195, 1973, p 48. V.M. Zinchenko, B.V.Georgievskaya, and V.v.
135-144 Kugnetsov, Improvements of the Technologi-
35. H. Brugger, "Impact-Bending Test for the As- cal Process of Automotive Parts Carburisation,
sessment of Case-Hardening Steels," conf. pa- Metallovedenie I Termi- cheskaya Obrabotka
per presented to the Swiss Materials Testing Metallov, August, 1979, p 47-50
36. J.P. Sheehan and MAH. Howes, 'The Effect
of Case Carbon Content and Heat Treatment SELECTED REFERENCES
on the Pitting Fatigue of 8620 Steel," Paper
720268, presented to the Society of Automo- • R. Charterjee-Fischer, Cause and Effect of In-
tive Engineers, 1972, p 16 ternal Oxidation, Antriebstechnik, Vol 25 (No.
37. D.E. Diesburg, C. Kim, and W. Fairhurst, 6), June 1986, p 41-43
"Microstructural and Residual Stress Effect • R. Chatterjee-Fischer and H. Kunst, Some
on the Fracture ofCase-Hardened Steels,"Paper Considerations on the Property Behaviour of
23, presented at Heat Treatment 1981, Sept Salt-Bath and Gas Carburized Specimens
1981 Made of Steel 16MnCr5, Hart.-Tech. Mitt.,
38. D.E. Diesburg and G.T. Eldis, "Fracture Vol 43 (No.1), JanlFeb 1988, p 41-44
Toughness and Fatigue Behaviour of Carbu- • JA Colwell and R.A. Rapp, Reactions of
rised Steels," Laboratory Report No. Fe-Cr and Ni-Cr Alloys in CO/C02 Gases at
L-193-87/88, Climax Molybdenum, 1975 850 and 950°C, Metall. Trans. A, Vol 17 (No.
39. D.E. Diesburg, High Cycle and Impact Fatigue 6), June 1986, p 1065-1074
ofCarburised Steels, SAE Paper 780771, 1978 • L. Faminski, M. Golebiowski, and J.
40. G. Fett, Bending Properties of Carburised Szawlowski, Influence of High Temperature
Steels, Adv. Mater. Process. inc. Metal Prog., Carburizing on the Structure and
April 1988, p 43-45 Properties of Steel Alloyed With Titanium,
41. T. Aida, H. Fujio, M. Nishikawa, and R. Inz: Materialowa, Vol 7 (No.6), NovlDec
Higashi, Influence of Impact Load on Fatigue 1986, p 14fr-152
Bending Strength of Case-Hardened Gears, • K. Isokawa and K. Namiki, Effects of Al-
Bull. Jpn. Soc. Mech. Eng., Vol 15 (No. 85), loying Elements on the Rotating Bending Fa-
tigue Properties of Carburized Steels, Denki
1972, P 877-883
Seiko (Electr. Fum. Steel), Vol 57 (No. 0, Jan
42. Y. Terauchi and J. Takehara, Paper 152-21,
1986, p 13-22
Bull. Jpn: Soc. Mech. Eng., Vol 21 (No. 152),
• H. Kanizawa and H. Satoh, Effect of Alloying
Feb 1978, p 324--332
Elements on the Formation of Internal Oxida-
43. S.E. Manevskii and I.I. Sokolov, Resistance to
tion Layer in Carburized Steels, Netsu Shari
Seizing of Carburised and Carbonitrided
(1 Jpn. Soc. Heat Treat.), Vol 37 (No.4),
Steels, Met. Sci. Heat Treat. (USSR) (No.4),
1997, P 225-230
April 1977
• M. Kikuchi, A. Komine, and Y. Kibayashi,
44. T.S. Eyre, The Mechanisms of Wear, Tribol. Influence of Internal Oxides on the Fatigue
Int., April 1978, P 91-96 Strength of Carburized Steel, I. Soc. Mater.
45. V.I. Astraschenko and L.A. Karginova, Spotty Sci., Jpn., Vol 38 (No. 425), Feb 1989, p
Carburising of Steels Metallovedenie I 111-116
Termicheskaya Obrabotka Metallov, June 82, p • Y.H. Kim and S.w. Lee, Internal Oxidation
13-15 Control of the Steels During Carburizing in
46.I.S. Kozlovskii, S.E. Manevskii, and Cracked Methanol Gas Atmospheres, J. Ko-
Kazachenko, Effect of Quenching Conditions rean Inst. Met. Maten, Vol 29 (No. 11), 1991,
on the Layer Structure and Antiseizing Stabil- P 1131-1138
ity of Carburised Steels, Metallovedenie I • T. Kimura and K. Namiki, Carburizing and
Termicheskaya Obrabotka Metallov, June Fatigue Properties of Plasma Carburized
1980, p 7-10 SCM 420, Denki Seiko (Electr. Fum. Steel),
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
36/ Carburizing: Microstructures and Properties

Vol 61 (No.1), Jan 1990, p 32-40 39 (No.6), June 1997, p 1021-1035

• J.Y. Liu and S.C. Chang, The Oxidation and • G. Prunel, Internal Oxidation During the Car-
Carburization of Fe-Mn-Al Alloys in a Car- burizing of Gears, Trait. Therm., Vol 197,
bon-Containing Atmoshere, Corros. Sci., Vol Dec 1985, P 31-41
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Carburizing: Microstructures and Properties Copyright © 1999 ASM International®
Geoffrey Parrish, p 37-49 All rights reserved.
DOI: 10.1361/cmap1999p037

Chapter 2


Decarburization, as the term implies, is a loss spheres for carburizing and reheating are carbon
of carbon atoms from the surface of the dioxide (CO z), water vapor (HzO), hydrogen
workpiece, thereby producing a surface with a (Hz), and oxygen (Oz). Under certain conditions,
lower carbon content than at some short distance these gaseous molecules can react with the car-
beneath the surface. If carburization promotes a bon atoms at the gas-metal interface and thereby
positive carbon gradient, then decarburization extract them from the surface of the metal. This
promotes a negative carbon gradient. extraction is an attempt to establish some mea-
The useful properties developed by carburiz- sure of equilibrium between gas and metal. The
ing and hardening will not be realized if the chemical reactions involved are:
working surface of the component becomes
decarburized. Therefore, decarburization is an
unwanted metallurgical feature. The optimal
amount of decarburization is regarded as being
zero, but in reality, a small amount will likely oc-
cur. A small amount of decarburization may be
tolerable if it is partial decarburization and pene-
trates no deeper than any HTTP associated with
the surface internal oxidation and it is within any
specification for decarburization. When these reactions proceed from left to
In well-run carburizing and hardening facili- right they are decarburizing, and when they go
ties, the decarburization of case-hardened sur- right to left they are carburizing. There seems to
faces does not seem to be a problem. This suc- be some disagreement regarding the third reac-
cess reflects the use of good, well-maintained tion (involving methane) and whether it is able to
equipment and good process operating procedures. proceed in both directions as reported in Ref 1. It
Nevertheless, with case hardening, the gas has been reported as being able to proceed in one
mixtures employednecessarily containdecarburizing direction only (Ref 2). If the latter is the case,
agents. As long as these agents are present, there then any hydrogen released by this reaction must
is always the possibility that a loss of furnace at- combine with other component gases before it
mosphere control, for whatever reason, will lead can be an effective decarburizing agent. By com-
to decarburization of the parts being treated. bining with any oxygen in the furnace atmo-
sphere, it will produce water vapor at the metal
surface and thereby enter into the second of the
Decarburization Processes
reactions shown, in a left to right direction.
Decarburizing reactions can occur at tempera- Decarburizing Conditions. During carburizing,
tures above about 700 °C and when, in the fur- there are numerous carburizing and decarburizing
nace atmosphere, decarburizing agents are avail- reactions occurring simultaneously, with more
able to react with the carbon in the metal surface. carburizing than decarburizing reactions taking
The decarburizing agents used in furnace atrno- place (i.e., the reactions proceed from right to
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
38 / Carburizing: Microstructures and Properties

left). Changes of temperature, flow rate, or com- words, the carbon potential will be reduced (Fig.
position of the atmosphere can shift the balance 2.1).
so that the number of decarburizing reactions Typical CO2 contents during a carburizing cy-
outstrips the number of carburizing reactions cle are shown in Fig. 2.2, where an endothermic
(i.e., the reactions proceed from left to right), and carrier is used throughout and is enriched for the
the overall result is decarburization. carburizing part of the operation. Normally, the
In an endothermic gas generator, for example, atmosphere at the end of the process, just prior to
if the catalyst is in good order, the methane of the the quench, will still be carburizing or at least of
output gas will be <2%. The carbon dioxide and a high carbon potential. If, however, the catalyst
water vapor will be present in only small is inefficient and, therefore, the generator gas
amounts, and these amounts are dependent on contains more CO2 than it should, then some
the actual air/fuel ratio employed. If, on the other decarburization is possible.
hand, the efficiency of the catalyst is impaired by When the reheat quench process is being used,
an accumulation of soot, for example, then the matching the atmosphere to the as-carburized
gas produced will contain greater amounts of surface carbon content might not be very precise,
methane, carbon dioxide, and water vapor (Ref 3). and a minor lowering or raising of the surface
This increase in the proportion of decarburizing carbon can occur. This practice is normally ac-
agents will lead to a reduction in the number of cepted.
carburizing reactions taking place at anyone Following are some reported examples regard-
time and, concurrently, an increase in the number ing the occurrence of decarburization during
of decarburizing reactions taking place. In other high-temperature heating.
Gutnov et al. (Ref 5) observed that samples
placed in a furnace at the beginning of an 800 -c
heating cycle decarburized appreciably, whereas
those introduced into the furnace halfway
through the heating period did not decarburize
(Fig. 2.3). Gutnov et al. concluded that moisture
in the furnace had been responsible for the dif-
2.0 t---1T-+---+-+-\-----\

1000 1h 2.5 h 1.5 h

V Presoa Carburize (935 ·C)
1.6 t---rl-\+I-H<-'~---;-+-----\ ~
:::J (935 ·C
e 900
E Atmoshpere: Equalize
~ 800 20.5% CO
tft. (800 ·C)
d' 1.2 I---\-lrt\-\I--\~----J~--t-\r------\ 40.0%H 2

O.60%C tft. 0.30
0.70% C -~'"
0.80% C -1---*-4.'\
0.4 I----+--T-~~-k-''''<
'" 0.20


o"-% CO2 + %....I-

CO = 20 ...L.. --'

700 800 900 1000 0

0 2 3 4 5 6
Temperature, ·C Time, h
Fig. 2.1 Temperature and percentage of carbon-
dioxide for equilibrium conditions with carbon Fi~. 2.2 Complex industrial carburizing cycle. Source:
steelsof various carbon contents. Source: Ref 1 ReT 4
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Decarburization / 39

ference. With an excess of moisture in the fur- scaling of an SAE 8620 steel after heating for 2 h
nace atmosphere (i.e., an increased dew point), in a nitrogen-4% natural gas atmosphere (Fig.
the carbon potential would fall in accordance 2.7). Such work underlines the importance of
with Fig. 2.4. equipment maintenance, in this instance with re-
Virta (Ref 6) presented the time-temperature spect to leaks and air ingress.
relationship for decarburization of a 0.9% tool Ambrus and Pellman (Ref 8), investigating
steel during reheating in air in a fluidized bed nitrogen-base atmospheres for the heat-treatment
furnace (Fig. 2.5). Thus, decarburization to a of 4340 and 300M steels, found that the three at-
depth of 200 urn (0.008 in.) was induced by heat- mospheres (pure nitrogen, methanol CAP-air,
ing in air at 900 °C for about 2 h. Kileeva et al. and methanol CAP-C02) produced some partial
(Ref 7), studying methods of coating to protect decarburization as well as some internal oxida-
against the decarburization of high-carbon sur- tion (Table 2.1). The interesting result of this
faces during normalizing at 890°C, observed work is that, for a given steel, the depth of
that the surface carbon would fall to under 0.4% decarburization was similar for all three atmo-
in 1 h in air at 900 °C (Fig. 2.6). spheres (300M steel, -69 urn; SAE 4330, -52
Stratton (Ref 2), whose work considered atmo- urn), whereas the internal oxidation was signifi-
sphere contamination in general, showed the in- cantly less when pure nitrogen was used.
fluence of air leakage onthe decarburization and Sagaradze and Malygina (Ref 9) showed that
heating in a salt bath (50% NaCl + 50% KCl)
caused decarburization. The effect became worse
Int?oduced halfwa as the salt bath aged, and deoxidation of the bath
through the 0 failed to overcome the problem.
• 0.8 soak at \
?J' 800 °C Introducedat the
g 0 start of the heat treatment
~ 411V1l'----------------.,

0.2 0.4 0.6 0.8 1.0 1.2

Depth. mm
Fig. 2.3 Decarburization due to moisture in the fur-
nace fireclay lining. Source: Ref 5

Dew point. OF
20 40 60 60 90 120
Timeat temperature. min

Fig. 2.5 Depth of decarburization of a cold-work steel

in a fluidized bed in air. Source: Ref 6

Upto 900 °C
'. ~950°C
' .
.: 800°C' ..~
~ 0.41----+--+-~~._+--__:_t--____j

0.1 ' - - - ' - - - ' - - -........................................._.u..J,----J - ..
-30 -20 -10 0 10 20 30 40
Dew point. °C o 15 30 60 90 120
Fig. 2.4 Variation of carbon potential with dew
point for an endothermic-based atmosphere con- Fig. 2.6 Decarburization as a function of holding
taining 20% CO and 40% H 2 in contact with time at different temperatures. Solid lines, samples
plain-carbon steel at various workpiece tempera- with protective coating; dashed Iines, samples with-
tures. Source: Ref 1 out protective coating. Source: Ref 7
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
40 I Carburizing: Microstructuresand Properties

The choice of atmosphere carbon potential for to that steel. Theoretically, an endothermic car-
reheating prior to quenching is most important. rier gas, perhaps with a slight enrichment (e.g.,
Naisong et al. (Ref 10), with a 52100 steel 0.5% C potential), should have been adequate.
(-1.02% C) and a reheating temperature of 810 The tests showed that to obtain the best results
°C, employed endothermic atmospheres with the atmosphere had to be carburizing. Anything
carbon potentials between 0.15 and 0.9% to de- less resulted in a reduced hardness and residual
termine which carbon potential was best suited tensile stresses at the surface (Table 2.2).
The Physical Metallurgy of Decarburization. If
a carburized steel is in the austenitic condition at
10 r - - - - , - - - - - - - - - - - - , - - - ,
,,, I
a temperature above the AC3 (e.g., 900 "C) and if
the furnace atmosphere is decarburizing (i.e., the
,, Variableresults
due to oxidation I
reactions proceed from left to right), then carbon
atoms will leave the surface in an endeavor to re-
store equilibrium with the surrounding furnace
atmosphere. A negative carbon gradient is pro-
6 Total duced with carbon atoms feeding down the gra-
decarburization dient. The surface carbon content is determined
.: by the carbon potential of the atmosphere. If the
« potential is low, the surface carbon content will
4 be correspondingly low. The depth of the
decarburized layer is determined by the length of
time that the part resides in the outlined furnace
conditions. At temperatures below the AC3 and
above the ACt, including those used for reheating
and quenching (800 to 840°C), the
decarburizing reaction is somewhat different
(Ref 11). At these temperatures, the carbon con-
0.1 0.2 0.3
Depth of carburization. mm tent of the immediate surface (originally A in
Fig. 2.7 Effect of leak rate on the decarburization of Fig. 2.8 and 2.9a) rapidly falls to the value
SAE 8620 after 2 h at 850°C in a nitrogen/4% natural gas shown by B. Any further lowering of the carbon
atmosphere. Source: Ref 2 content must result in a material of carbon con-

Table 2.1 Laboratory results: effectofatmospheres on high-strength steels

Depth ofdecarhurizalion."m (ilL) Deplh ofinlergranular altack,,,m(jJilL) Average lensile s1renglh,
Almosphere Average Siandarddevialion Average Siandarddevialion MPa(ksl)
Nitrogen 48(0.0019) 20.3 (0.00080) 0.5 (20) 0.5 (20) 1627(236)
MethanolCAP-air 58(0.0023) 7.9 (0.00031) None visible None visible 1658(240.5)
MethanolCAP-CO2 51 (0.0020) 7.9 (0.00031) 11.2 (440) 1.0 (40) 1612(233.9)
Nitrogen 66(0.0026) 19.8 (0.00078) 1.0 (40) 2.5 (100) 2013 (292)
MethanolCAP-air 69(0.0027) 0 4.1 (160) 1.0(40) 2015 (292.3)
Methanol CAP-CO2 69(0.0027) 0 5.8 (230) 0.8(30) 2025 (293.7)

Source:Ref 8

Table2.2 Effect of carbon potential on microstructure, hardness, and residual stresses ofa 52100steel
(1.02%C)duringreheating priorto quenching
Atmospherecarbon potentialzeyo Relained auslenite, % Carbides, % Hardness(a), HK Surface residualslresses, MPa
0.9 17.4 16 820 -40
0.7 9.8 12 811 +10
0.5 6.2 8.5 751 +140
0.3 2.5 7 695 +270
0.15 2.0 7 626 >270

Reheated 10 810 DC for I h, oil quenched at 55 DC, tempered at 150DC for 2 b. HK, Knoop hardnessnumber.(a) I kg load. Source:ReflO
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Decarburization /41

tent C being formed in equilibrium with material When a controlled atmosphere is employed
of carbon content B. Therefore, because the at- with a carbon potential of some value between A
mosphere is decidedly decarburizing in nature, a and B (Fig. 2.8), such as D, then a ferrite layer
further lowering of the average surface carbon cannot form, Instead, a gradient is produced be-
content must result in the development of ferrite tween carbon contents A and D (curve 1, Fig.
containing carbon to the equilibrium value C 2.9b). With time the gradient becomes flatter
(curve 2, Fig. 2.9a). This layer, once it becomes (curves 2 and 3, Fig. 2.9b) until eventually no
continuous, reduces the rate of decarburization gradient to the surface exists (curve 4, Fig. 2.9b).
because, in the low-carbon ferrite layer, the car- The two-stage carburizing method utilizes this
bon activity is reduced. Further, because the fer- gradient depletion process. In the first stage, a
rite can have only a shallow gradient across it, high carbon potential is employed to produce a
the rate of flow of carbon atoms through the high-carbon surface (e.g., 1.2% C). For the sec-
layer is also reduced. As the ferrite layer in- ond stage, the carbon potential is reduced to pro-
creases in thickness, the decarburizing rate is duce the specified surface carbon content (e.g.,
further reduced due to the decreasing effective- 0.85% C). Carbon diffuses both inward and out-
ness of the driving force responsible for moving ward, and the end result is a carbon gradient with
carbon atoms through the layer. Behind the fer- a surface carbon plateau at 0.85%. Too short a
rite layer, the carbon gradient in the austenite be- duration or too high a CO 2 content for the sec-
comes more gradual with time as carbon feeds ond stage results in a carbon gradient with a neg-
down to the interface with the ferrite layer (curve ative slope at the surface (Fig. 2.10). A negative
3, Fig. 2.9a). carbon gradient formed in this way does not lead
to a rejection, provided the surface carbon con-
tent and the surface hardness are each within
their specified range. Nevertheless, detection of
this problem should lead to a reassessment of ap-
propriate process parameters. A negative carbon
gradient of this type could produce an acceptable
martensitic microstructure at the surface with a
subsurface microstructure that contains excessive
a amounts of carbides or retained austenite.
Whether or not such a combination of micro-
E structures is accepted depends on the applica-

Carbon, % Testing for Decarburization. Heat-treatment
Fig. 2.8 Iron-carbon equilibrium diagram used to ex- operators go to great pains to avoid classical
plain decarburization decarburization because of its adverse effects. Its


CL...- -I C.... ---J

Depthbelow surface Depthbelow surface

Fig. 2.9 Effect of carbon potential on the surface carbon of a

decarburized surface. (a) Atmosphere decarburizing: carbon po-
tentialless than B.(bl Atmosphere decarburizing: carbon potential
equals D
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
42 / Carburizing: Microstructures and Properties

detection, through macrohardness and/or micro- HTIP only. With still more surface carbon (say,
hardness testing, metallography, carbon analysis, 0.4-0.5%), the microstructure of the decar-
or all three, will lead to the need to rework or burized layer might be martensitic. However, this
more probably the rejection of the affected com- martensite, having a low carbon content, will
ponents. An occurrence of decarburization might nital etch to be more gray in color than would a
be suspected if something untoward happened high-carbon martensite, and, therefore, this level
during the heat-treatment sequence. Such a sus- of decarburization should be detectable metallo-
picion precipitates action to determine if the graphically. For a given loss of surface carbon,
heat-treated parts suffered as a result of the oper- the final microstructure will depend on the alloy
ating problem. content of the steel and the cooling rate. (See Fig.
Soft surfaces, determined during routine qual- 1.13a and b and Fig. 1.14 for examples.)
ity testing of the components (by file or hardness
test) or of test pieces that accompany the furnace
load, will only indicate that something is wrong.
Soft surfaces can occur due to high-temperature
transformation products (HTIP) that accompany
internal oxidation, excessive retained austenite,
nonmartensitic structures resulting from either a
"slack" quench or a steel with an uncharacteristi-
cally low hardenability (extremely fine-grained),
or decarburization. The first action taken must
determine exactly which of the possibilities
caused the soft skin effect.
The Metallography of Decarburization. Decar-
burization is classified as either total or partial.
Total decarburization refers to the removal of es-
sentially all the carbon from the immediate sur-
face; the polished and etched surface micro-
structure is ferritic (Fig. 2. 11a). It will likely (a)
have a layer of partial decarburization beneath it.
Partial decarburization is any surface carbon loss
that is not sufficient to produce total decar-
burization (a condition somewhere between
curves 1 and 2 in Fig. 2.9a). Many consider par-
tial decarburization to be that which occurs when
metallography detects HTIP and grain-boundary
ferrite at the surface of the part (Fig. 2.11b). For
such a microstructure, the surface carbon content
would be quite low. With a little more surface
carbon, the microstructure might consist of

0.9 r--;;-.,.-~"...-,...---'------'----,---,
3 EN32A (0.029%AI)
0.8 ~~~t:----r----=-T4=---r-------j
C 0.5 ~-+---t--__t_"'''''-:----f''Iho:-+-_+-_____1
o Carburizingat 925 'C
~ 0.4
1 For 16 h at 0.16% CO
U 0.3 2 For 8 h at 0.1%C0 21.25h at 0.175% C02:-'l"~~
0.2 3 For 8 h at 0.1% CO2 +2.5 h at 0.175%CO2 --,:--t-;;----,;----t
4 For 8 h atO.1% CO2 +4 h at 0.175% CO2
5 For8hatO.1%C02+1 hatO.25%C02
Fig. 2.11 Micrographs showing different degrees of
0!:---:-:-:::--"'~--=-=---=,"=---;'-;:----7:=----;' decarburization. (a) Total decarburization caused by se-
o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 vere furnace leak during gas carburization of 1018 steel
Depth belowsurface. mm (l % nital etch, 500x). (b) Partially decarburized speci-
Fig. 2.10 Typesof carbon profiles. Source: Ref 12 men. 190x
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Decarburization / 43

The surfaces of quenched parts that have suf- an instance, the hardness could be either in-
fered partial decarburization will likely be creased or decreased by some small, and
bainitic or even martensitic depending on the al- possibly insignificant, amount. Hardness in-
loy content. A bainitic structure at the surface is creases are obtained when decarburization is re-
distinct compared to the underlying martensite. sponsible for lowering the retained austenite
Although in a well tempered structure (dark etch- content consequently increasing in the
ing), it might not be too apparent at first sight. martensite produced; the carbon gradient for this
Low- to medium-carbon martensites in a quenched is similar to the upper curve of Fig. 2.12(b).
and tempered surface (through the microscope) ap- Hardness reductions are obtained when low-car-
pearto be more gray than the higher-carbon mate- bon martensites or bainites are produced from a
rial beneath the affected layer. carbon gradient such as Fig. 2.12(a). If the re-
Any indication that a surface is decarburized duced hardness resembles a "skin" effect, it
justifies a microhardness traverse being carried might not be detected by macrohardness testing.
out. Small amounts of decarburization might not Microhardness testing, however, is expected to
be too obvious, metallographically or in terms of determine the presence of such microstructures.
a microhardness traverse. However, a good dense For the relationship between hardness and car-
martensitic surface layer with free carbides or bon content for the different microstructures, see
fair amounts of retained austenite beneath it is an Fig. 1.18.
indicator of a negative carbon gradient (upper With severe decarburization, even if proeutec-
curve, Fig. 2.12b). It is then necessary to conduct toid ferrite is not formed, appreciable amounts of
chemical analysis to confirm any suspicions of other low-carbon transformation products occur
this condition. at the surface. This presence of decarburization
can be detected by macrohardness testing. How-
ever, it should be remembered that surface soft-
Influence on Material Properties ness can also be caused by high levels of re-
tained austenite or excessive amounts of H1TP
Influence on Hardness. Shallow decarburization, associated with internal oxidation, in which case
or a minor reduction of the surface carbon con- it is necessary to utilize metallography to deter-
tent, does not greatly influence the surface mine the cause. Having said that, for a given
macrohardness of a case-hardened part. In such depth of H1TP associated with internal oxida-


Distance from surface Distance from surface Distance from surface

B +
> 800
:ll 700

500'-- -'
Distance from surface Distance from surface Distance from surface
(a) (b) (e)
Fig. 2.12 Influence of decarburization on hardness profiles. F, ferrite; B, bainite; M, martensite; A, aus-
tenite. (a) Severe decarburization. (b) Slight decarburization. (c) No decarburization
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
44 / Carburizing: Microstructures and Properties

tion and the same depth of HTTP due to Influence on Residual Stresses. When a carbu-
decarburization, the effect on other properties rized surface without decarburization is
might not be all that different. quenched, transformation progresses inward then
outward from the case/core interface (Fig. 2.13).
When a decarburized component is quenched
and some high-temperature transformation prod-
Development of structure during quench uct is formed at the surface, transformation (be-
- - - - - Temperature,·C ----n sides progressing inward then outward from the
case/core junction) will take place prematurely at
DIlll Carbide ~ Pea~lte the surface (Fig. 2.14). The decarburized layer,
Austenite Bainije
in this example, contains almost 50% bainite,
590 lIIIJ Ferrite Martenslte 626 808
and the volume expansion is less than that of the
underlying martensite. Consequently, when the
504 694
474 whole of the case has eventually transformed, the
'----_--=rrfriJJJJIWJL surface will be in less compression, or even in
tension, compared with a fully martensitic
An example of the influence of decar-
burization on residual stresses within a carbu-
rized surface is shown in Fig. 2.15. With no
decarburization present and a surface carbon
content of about 1%, the residual compressive
stresses at the surface are in excess of 392 MPa
(40 kg/mm 2). When sufficient decarburization
occurs to lower the surface carbon content to
0.64% and to penetrate to an estimated depth of
0.3 mm, the residual stresses at the surface are
virtually zero. When decarburization causes the
surface carbon content to fall to 0.35% and the
decarburized layer is about 0.5 mm deep, the re-
sidual stresses at the surface are tensile to 226
MPa (23 kg/mm-). Tests on the decarburization
of a 52100 bearing steel (Ref 10) suggest that

1200 1.2

1000 1.0

800 0.8
s: - ~"'"
~ .-
Final structure at ambient (30 ·C) ~600
r- :r'"
400 0.4

~ '\:
30 Jo 3l I"A-. Ha~ness

T~ 1'1
. , 200 I 0.2

2 3 4
o o
o 2 3 4 5
Distance from surface, mm
Distance from surface, mm
(a) (b)

Fig. 2.13 Progression of transformation in carburized EN36A (655M 13) steel. (a) Transformation occurs first in the
core region before any martensitic transformation in the case region. The final structure in the case is predominantly
martensitic with some retained austenite. Ruling section: 125 mm; heat treatment: oil quenched from 820°C; case car-
bon: 0.8% at surface. (b) Carbon and hardness profiles. Source: Ref 13
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Decarburization / 45

only small reductions of surface carbon are suffi- surface (Table 2.2). These examples imply that
cient to produce tensile residual stresses at the through hardened high-carbon surfaces are per-
haps more sensitive to small amounts of
Development of structure during quench
decarburization than are carburized surfaces.
Influence on Bending Fatigue Strength. A con-
- - - - - Temperature, DC ----ll sequence of decarburization and of reducing sur-
IIIIIl Carbide ~ Pearlite face hardness and developing unfavorable resid-
Austenite Bainite ual stresses is to impair the bending fatigue
642 [Ill Ferrite Martensite 662 nl strength of the components. Gunnerson (Ref 14),
using rotating bending fatigue specimens, found
568 that decarburization and heavy internal oxidation
585698 reduced the fatigue limit from -81 to 53 kg/mm-,
dililllilllllilillililililllilililiifIIl] Reductions of fatigue limit exceeding 50% were
obtained by decarburizing the surface of a
case-hardened Cr-Mn-TI steel to 41 to 42 HRC
(Fig. 2.16) (Ref 15). Figure 2.16 also shows that
decarburization can nullify the benefits of carbu-
rizing by reducing the fatigue limit to approxi-
mately that of a core containing approximately
0.19% C with a hardness of 30 HRC. Sagaradze
and Malygina (Ref 9), with a 20Kh2N4A steel,
obtained -50% reduction in bending fatigue by
reducing the surface carbon from about 0.8 to
less than 0.35%; the respective fatigue values
were 784 MPa (80 kg/mm-) and 353 MPa (36
kglmm2) . The corresponding residual stress dis-
tributions are presented in Fig. 2.15. Fatigue

1000 1.0

800 ~ 0.8
0.6 <f


Final structure at ambient (30 DC)

31 Hardness

lJ Jl 200 0.2
h- n.lli. Carbon

o o
0.5 1.0 1.5 2.0 o 2 3
Distance from surface, mm
Distance from surface, mm
(a) (b)

Fig. 2.14 Transformation sequence in decarburized EN34 (665M17) steel. (a) Surface decarburization causes trans-
formation to initiate at the surface in addition to at the case/core interface. This effect reduces not only surface hardness
by the presence of bainite, but it also reduces the level of surface residual compressive stress (possibly even into the ten-
sile range) by the change in transformation sequence. Predictions are based on the following input data: ruling section,
100 mm; heat treatment, austenitized at 925°C and direct oil quenched from 820 °C; case carbon, 0.8% falling to 0.5%
at the surface. (b) Carbon and hardness profiles. Source: Ref 13
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
46 / Carburizing: Microstructures and Properties

tests, using notched test bars that simulated a 13 decarburized surface would be detrimental to the
mm modulus gear (-2 diametrical pitch), contact durability of a carburized and hardened
showed that the fatigue limit was reduced by al- part. In the absence of test data, it is impossible
most 40% on those samples decarburized to a to determine what would happen if a
depth of 0.22 mm (Fig. 2.17). Kileeva et al. re- decarburized part were to enter service. Natu-
ported reductions of fatigue strengths of 19% in rally, it greatly depends on how much rolling and
one case and 35% in another case from the decar- sliding are involved, what the contact pressure is,
burization that occurred during post-carburizing and how severe the decarburization is.
normalizing or hardening heat treatments (Ref Consider a SAE 8620 part manufactured to
7). The microstructure of the decarburized layer function as a case-hardened part, which has a
has the most influence on the fatigue strength; decarburized surface carbon content of only
the depth of decarburization (within reason) and 0.45% and a penetration depth corresponding to
the microstructure of the underlying material are maximum shear. Early contact cycles deform the
probably not very significant. softer surface layers and thereby quickly contrib-
Influence on Contact Fatigue. Contact fatigue ute to a good load distribution. At the same time,
resistance is related to the shear strength of the the soft material work hardens to improve its re-
material; therefore, it would be expected that a sistance to wear. However, it can only be de-
formed so much before it becomes overworked
and the amount of spreading gives rise to ex-
40 treme tensile residual stresses. These stresses oc-
3.5Ni-l.5Cr I~
30 cur in a narrow zone somewhere within the
/ I.-
200 8!
decarburized layer and possibly close to the in-
terface with harder underlying carburized mate-
fI.l 0
1\ / r..J
o i rial. The outcome of that circumstance is surface
spalling, which involves flakes of work hardened

""' \ 1\ ' ~
v .-/
surface being detached from the surface. The
time to failure would not be long.
If the amount of decarburization was confined
1\ ~ V
-30 to the outermost 75 urn, for example, then the
/ Case adverse effects of decarburization might not be
~ ........ V depth --= ~OO
-50 .1 great. See the corresponding section in Chapter
o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 1, "Internal Oxidation," with attention to the
Fig. 2.15 Effect of decarburization on the residual work of Sheehan and Howes.
stresses developed in carburized and hardened plates.
The carbon content at 0.002 mm was estimated to be 1%
(curve 1),0.64% (curve 2), and 0.35% (curve 3). Source: 800 ...-........- ......----...---.--....---.--.----.
Ref 9 ~ Decarburtzatlon free EN36
~ 700 - ......_~ Total case 2.5 mm
• 600
" Decarburtzed

500 :
70 0 Carburized, 1.2 mmcase -v-_ _..ll800 .~ 400 I
D Decarburized to :0 Core
41-42 HRC at surface 300 L_-'----'_.........._ _ ' _ - - 0 . _........_ " ' -.......
600 0 0.5 1 1.5 2
• Not carburized-numbers
in parentheses give (SO)• •
(8) Distancefrom surface,mm
90 .-----r-~-___r----,r_:_;_-:-::-:-"l14oo
approximate HRC Ki= 1.55
• •
30 •
(38) (45)
~ 70
g 60 1000 I~
150 800
20 -(12)

0.1 0.2 0.3 0.4 30 1()3 104 105 106 107

(b) Cycles
Core carbon, %
Fig. 2.16 Effect of decarburization on the bending fa- Fig. 2.17 Effect of decarburization in carburized
tigue strength of a Cr-Mn-Ti steel of varying core carbon. EN36 0.16% Ni, 0.9% Cr) steel on (a) microhardness and
Source: Ref 15 (b) bending fatigue strength
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Decarburization I 47

Influence on Bending and Impact. Decar- Rectification of components once they are
burization does not necessarily reduce the static found to be decarburized depends on a number
bending strength of case-hardened surfaces (Ref of factors. Questions must be asked and an-
9). As Table 2.3 shows, bend strength peaks swered. For example, does the amount of decar-
when the surface carbon content is 0.6%, despite burization measured render the part useless? Is
the almost zero residual stresses at the surface. the item to be stressed to its undecarburized limit
Presumably, this corresponds to the highest car- in service, or is there a fair safety margin built
bon content that produced only martensite. into the design? If the decarburization is shallow,
Influence on Wear. The adhesive and abrasive can it be grit blasted in the critical areas to re-
wear resistance of a case-hardened surface is ad- move the affected layer and shot peened to en-
versely affected by decarburization to a degree sure surface compression? If the decarburization
relating to the surface hardness (see Fig. 1.27). is deeper than what can be reasonably grit blasted
and shot peened, can the part be recovered by
reclamation heat treatment (Ref 16)? It must be
kept in mind that any reheat treatment could in-
Control of Decarburization
duce additional shape and size changes (distor-
tions) that on their own could warrant a rejection
Severe decarburization of case-hardened parts
of the part.
occurs only rarely, especially with modem atmo- Each incident of decarburization must be as-
sphere monitoring systems and when furnaces sessed on its own; what might work for one part
can be nitrogen purged at the first signs of trou- design might not be applicable to another. The
ble. Even improvements in health and safety economics of encountering decarburization is
monitoring mean safer handling of parts, so there obviously an important consideration. Whatever
is less risk of accidents occurring during transfer action is taken to recover the situation, the pri-
of parts from furnace to quench. The risk of en- mary consideration should be with respect to the
countering decarburization is further reduced by component and its eventual fitness for service.
having good plant maintenance, good supervi-
sion, and sound process control.
The less obvious type of decarburization (i.e.,
caused by too short a diffusion stage or an incor- Summary
rect atmosphere adjustment in the boost/diffuse
method) is fairly easy to correct. It is important
to be able to detect that there is a problem. Decarburization can occur when surface car-
bon is slightly lower than at depth due to an in-
correct diffuse stage of a boost-diffuse carburiza-
tion program. Surface carbon is adequate if
Table 2.3 Effectof decarburization on the fatigue acceptable hardness is achieved. This condition
and bending strength of case-hardened 20Kh2N4A is not serious. When surface carbon is moderate
steel to very low, decarburization is detectable by
Surfacecontent, Fatiguestrell&th Bendstrength hardness test and by metallographic methods.
% (atmosphere) (o-I),MPa (0 hend),MPa
This condition is unacceptable, as it may result
in a failure.
1.04 (not normalized) 80 319
0.8 (gas furnace) 81 355 • Preprocess considerations: Good mainte-
<0.8 (nitrate salt) 77 318
0.6 (chloride bath. 20 min) 60 407
nance of equipment is essential. Decarbu-
0.35% (chloride bath. 160 min) 39 318 rization might signify an equipment failure.
Not normallzed(b) Long reheating times in air are not recom-
0.16 45 276 mended.
0.51 50 343 • In-process considerations: Atmosphere moni-
0.63 49 412
0.84 57 350
toring system might indicate a problem. Re-
1.05 51 307 heating in air is not recommended.
• Postprocess considerations: Post-process cor-
(a) Carburized, normalized at 890 °C. high-temperature tempered and
quenched from 800 "C, then tempered to 140 "C. (b) Carburized,
rections depend on the degree of decarbur-
high-temperature tempered. heated in chloride bath for various times ization. For shallow partial decarburization,
to decarburize, oil quenched at 800 °C, then tempered to 140 °C. consider shot-peening. For deeper total or
Source: Ref9
partial decarburization, consider restoration
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
48 // Carburizing: Microstructures
48 Microstructures and
and Properties

carburizing ifif added distortion can be toler- tential
tential on the Development
Development of of Residual
ated. Stresses in 52100 Bearing Steel, Metall. Trans.
Metall. Trans.
• Effect on properties:
Effecton properties:Significant
decarburization A, Vol 15,
15, Nov 1984, p 2101-2102
leads to incorrect
incorrect surface microstructures
microstructures and 11.
11. W.A. Mechanism of the Surface
WA. Pennington, A Mechanism Surface
low hardness values. If If surface carbon
carbon is Decarburisation
Decarburisation of Steel, Trans.
Trans. ASM,
ASM, 1946,
greater than 0.6%, the surface hardness Vol
Vol 37, p 48-91
should be acceptable. If If surface carbon is ap- 12. R. Collin, Mathematical Model
Model for Predicting
proximately 0.6% or less, all the main proper- Carbon Concentration Profiles, J. IronIron Steel
ties will be adversely affected,
affected, for example, Inst,
Inst., Oct
Oct 1972
bending fatigue could be reduced by 50%. 13. D.W.
D.W Ingham and P.C. Clarke, Carburise Case
• Standards:
Standards: ANSJlAGMA
speci- Hardening—Computer
Hardening-Computer Prediction of Structure
fication for grade 1.
I. For grade 2 and grade 3, and Hardness Distribution, HeatHeat Treat. Met,
Treat. Met.,
no partial decarburization
decarburization is apparent in outer 1983, Vol
Vol 4, p 91-98
0.13 mm (0.005 in.) at 400x 400>< except in un- 14. S. Gunnerson, Structure Anomalies
Anomalies in the Sur-
ground roots.
TOOts. IS06336-5.2: for MQ and ME face Zone of Gas-Carburised Case-Hardening
grades, the reduction of of surface hardness due Met. Treat.
Steels, Met. Treat. Drop
to decarburization in the outer 0.1OJ mm (0.004 213), June 1963, p 219-229
in.) should not exceed 2 HRC on the test bar. Sagaradze, Effect of Carbon Content
15. V.S. Sagaradze, Content on
the Strength ofCarburised Met. Sci.
of Carburised Steel, Met. Sci.Heat
(USSR), (No.
Treat. (USSR), 3), March 1970, p
V.G. Koroshailov, E.L.
16. Y.G. E.L. Gyulikhandanov,
GyuUkhandanov, and and
1. Heat Treating,
Handbook,9th 9th V.V.
Y.Y. Kislenkov, Restoration Carburising of of
ed., American Society for Metals, 1981 Steel ShKhl5 in Endothermic Atmosphere,
2. P.E Stratton, Living with Furnace Atmosphere Met. Sci.
Met. Sci. Heat Treat.
Treat. (USSR),
(No. 12),
Treat. Met.,
Contamination, Heat Treat. Met.,1984,
1984,VolVol2,2, 1975, pP 33-35
3. L.H. Fairbank and L.G.W L.G.W. Palethorpe, "Con-
trolled Atmospheres for the Heat Treatment of of
Metals," Heat Treatment
of Metals,Special
port 95, Iron and Steel Institute, 1966, pp 57-69
4. FA Still and H.C. Child, Predicting Carbu- • P.
P. Baldo
Baldo and E. Duchateau,
Duchateau, "Process for HeatHeat
rising Data, Heat Treat.
Treat. Met;
Met., 1978,
Vol3,3,pp Treatment Under a Gaseous Atmosphere
67-72 Containing Nitrogen and Hydrocarbon,"
Hydrocarbon," U.S.
5.R.B. Gutnov, L.P. L.P. Emel'yanenko, E.N. Patent US4992113, 22 Aug 1990
Samsonova, L.A. Shvartsman, and D.D. M.J. Gildersleeve, Relationship Between De-
• MJ.
Shishlov, Sources of of Decarburisation during carburisation and Fatigue Strength of Through
the Restoration Case-Hardening of Steel in Hardened and Carburising Steels, Mater. Sci.
Mater. Sci.
Bell Furnaces, Stahl, (No. 11), Nov 1971, 1971, p Technol.,
Technol, Vol
(No. 4),April
1039-1040 H.J. Grabke,
• H.I. Grabke, E.M.
E.M. Muller,
Muller, H.Y.
H.V. Speck,
Speck, and
and G.
Virta, Heat Treatment '84 (London), The
6. The Konczos, Kinetics of the Carburization of
Metals Society, 1984 Iron Alloys in Methane-Hydrogen MixturesMixtures
7. A.1.
A.I. Kileeva, D.1.D.I. Potoskveva, and V.S. Hydrogen, Steel
and of the Decarburization in Hydrogen,
Sagaradze, Prevention of Decarburisation with (No. 5), 1985, p 275-282
Res., Vol 56 (No.5),
Enamel Coat, Met. Sci. Heat Treat. Treat. (USSR),
(USSR), B. Korousic and B.
• B. B. Stupnisek, Predicting of
(No. 10), Oct 1975, p 68-70 Reactions during Carburization and Decarbur-
8. Z.Z. Ambrus
Ambrus andand MA
M.A. Pellman,
Pellman, Hardening
Hardening of Atmospheres,
ization of Steels in Controlled Atmospheres,
Aerospace Alloys in Nitrogen Based Atmo- KovineZlitine
Kovine ZlitineTehnol.,
3030 (No. 6),
(No.6), Nov/Dec
spheres, Met. Prog.,
47-51 1996, p 521-526
V.S. Sagaradze
9. Y.S. Sagaradze and
and L.Y.
L.V. Malygina,
Malygina, Influence
Influence • G.
G. Matamala
Matamala andand P.
P. Canete,
Canete, Carburization
Carburization and
of Decarburisation on on the
the Properties of Decarburization Kinetics of Iron in 04-H CH4-H22
Case-Hardened Steel, Metal Sci. Sci. Heat Treat.
Treat. 1000-1100 "C,
Mixtures between 1000--1100 °C, Mater.
(USSR), (No.7),
(USSR), (No. 7), July 1966, p 560-563
560-563 Phys.,Vol
Phys., Vol1212(No.4),
(No. 4),April
10. X.X. Naisong,
Naisong, CAC.A. Stickels,
Stickels, and
and C.R.
C.R. Peters,
Peters, M. Renowden,
• M. Renowden, O. O. Borodulin,
Borodulin, and
and P.P. Baldo,
The Effect of Furnace
The Furnace Atmosphere Carbon Po- Unique Atmosphere Control System for Carbon
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Decarburization I 49

Control, Con! Proc. 61sf Annual Convention; Analysis of the Effects of the HeatingCharac-
1991 Regional Meeting (Atlanta, Georgia), Wire teristics on Steel Decarburizing, Russ. Metall.,
Association International, 1991, p 191-195 Vol 4, p 192-199
• G. Sobe and V. Polei,On the Reaction Mecha- • AF. Zhomyak, and V.E. Oliker, Kinetics and
nism of the Carburization and Decarburization Thermodynamic Characteristics of the Re-
of Alpha-Iron in C141H2-Mixtures, SteelRes., duction-Decarburization-Carburization of an
Vo157 (No. 12),Dec 1986, p 664-670 Atomized Cast Iron Powderin Converted Gas,
• B.S. Soroka. B.I.Bondarenko, AE. Erinov, V.E. Sov. Powder Metall. Met. Ceram; Vol 23
Nikolskii, and V. K. Bezuglyi, Thennodynamic (No.2), Feb 1984,p 102-107
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Carburizing: Microstructures and Properties Copyright © 1999 ASM International®
Geoffrey Parrish, p 77-97 All rights reserved.
DOI: 10.1361/cmap1999p077

Chapter 4

Retained Austenite

In steels, austenite is stable at temperatures Austenite Formation

above the AC3 and ACcm phase boundaries (see
Fig. 3.2). On cooling from such temperatures, it Austenite Stabilization. Retained austenite can
becomes unstable and decomposes to some new become stable. If a part with a high-carbon sur-
constituent depending on the chemical composi- face layer is quenched into a refrigerant (i.e., it is
tion of the steel and the rate of cooling. When the cooled straight through the M, to Mfrange), vir-
transformation involves diffusion processes (i.e., tually all the austenite will be transformed to
processes to form ferrite, pearlite, or bainite), the martensite. Conversely, if a part is quenched to
reaction is essentially complete and no austenite about room temperature, held there for some
survives. These resulting products are referred to time, and then refrigerated to below the Mf,
as high-temperature transformation products be- some of the austenite will transform isothermally
cause they form at relatively high temperatures. to martensite and some will survive. This surviv-
For example, for low-carbon steels, these trans- ing austenite is referred to as thermally stabilized
formations take place at temperatures between austenite, and it requires a fair amount of energy
the AC3 and about 400 °C. Martensite, on the to destabilize it.
other hand, is a low-temperature transformation Thermal stabilization involves a strain aging
product. For a typical low-carbon lean-alloy process (Ref 3), where the strain is provided by
steel, the martensite transformation range is the accommodation of the martensite and the
from 450 to 200 °C, whereas a high-carbon ma- presence of any tensile residual stresses. Stabili-
terial typical of a case-hardened surface has a zation requires the presence of interstitial atoms
martensite transformation range from 200 °C (e.g., carbon) and sufficient time for these at-
down to about -100°C. If, on quenching, part oms to segregate to dislocations or to the
of the martensite transformation range lies be- martensite embryo/austenite interfaces, thereby
low the temperature of the quenchant, the trans- pinning them (Ref 4). Once the segregation and
formation of austenite to martensite will remain dislocation pinning have taken place, the aus-
incomplete; austenite will be retained in the tenite is stiffened somewhat, and the growth of
final microstructure. Figure 4.1 represents a martensite is inhibited. In high-carbon austenite
high-carbon surface where a part of the typical of a carburized surface, interstitial atom
martensite transformation range (M, to Mf) lies segregation takes place rapidly. Whereas stabili-
below 20°C. Therefore, the presence of retained zation becomes more complete due to an iso-
austenite is to be expected and indeed is shown thermal hold, a slow cool through the M, may,
in Fig. 4.1(b). However, small amounts of re- nevertheless, be adequate for some stabilization
tained austenite have been detected in quenched to occur. In Fig. 4.1, the most rapid cool pro-
steels even when their M f temperatures are above duced an austenite content of 50%, whereas a
ambient (Fig. 4.2). Such austenite tends to reside slower cool produced 60%.
at interlath boundaries rather than as volumes The permanence of stabilization depends on
typical of a higher-carbon plate martensite. the extent to which the strain aging process has
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
78 I Carburizing: Microstructures and Properties

1000 ,.---------------;r;:--,
em 1700
800 1500

P 700 1300
~600 1100 ~
~ 500 900 ~ Q)

~ 300 ~
100 ~--_.:__-~~~
OL..._......I..:':O:::"_~::..-...L>ll_:__"Wl-~::::-_=:'----:! 100
1 103 106
(8) Time,s

;f. 120 r-~:::r.::;;::::::_------------,

,----, p 700 ~

80 "~ ,/ \, .i
600 ~


.~--_ ..-.'-""~~ 1\
t 500
400 ~
0; 20 RA = Residual austenite " I \ "E
g C / ••~ '\ ~
:! 0
1 10 1()2 103 104
Cooling time to 500 'C (930 'F), s

Fig. 4.1 CCl diagram with related hardness and percentage of

microstructural constituents for a 20Ni-Mo-6Cr steel carburized to
1.14% C. Austenitized at 930°C for 30 min. Source: Ref 1

been allowed to proceed and on what subsequent more complete degree of stabilization. Refrigera-
thermal (heating or cooling) or mechanical ac- tion, while transforming some austenite to
tions are brought to bear. The clusters of intersti- martensite, also makes any remaining austenite
tial atoms can either be dispersed or encouraged more ready to transform into bainite during
to precipitate (overaging). low-temperature tempering.
Mechanical working can cause some desta- The Relationship between Ms and Retained
bilization by inducing further transformation to Austenite, A consequence of developing a carbon
martensite, and tempering at temperatures above gradient in the surface of a steel component is
about 150°C can also destabilize austenite by that the M, falls as the carbon content increases.
transforming it to bainite. Room temperature ag- An indication of the efficiency of carbon to mod-
ing and tempering below about 150°C favors a ify the M, can be obtained from several sources,
but here the Steven and Haynes formula (Ref 5),
20'---------------~ which determines the M, from the chemical
composition of a steel, will be used:

M, (0C) = 561 - 474C - 33Mn - 17Ni


This formula is reasonably accurate for steels con-

taining up to 0.5% C. In higher carbon contents,
the efficiency of carbon to lower the M, is reduced,
O~_~~~~~~~ and the correction curves ofFig. 4.3 are required to
o 0.2 0.6
Carbon, % obtain a truer M, value. If this complication is ig-
Fig. 4.2 Influence of carbon content on retained aus- nored for the moment, the elements in the formula
tenite content. Source: Ref2 do lower the M, according to their quantity in the
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite / 79

steelandthe respective factor. For example, the fac- austenite (Vy) is therefore related to both the M,
tor for carbon is 474; therefore, a 0.1% C increase and the quenchant temperature (Tq). This rela-
lowers the M, by 47°C (progressively less as the tionship, deftned by Koistinen and Marburger
carbonlevelincreases above0.5%).For alloying el- (Ref 6), is shown in Fig. 4.5:
ements to have the same effectrequires an increase
of 1.5%Mn, nearly3% Ni or Cr,or slightly over2%
Mo. 'Therefore, carbon up to about 1% is much
more influential for lowering the M, through the
case than other alloying elementstypical of the car- Thus, methods are available whereby the M,
burizing gradesof steel. Figure4.4 showsthe effect can be determined from chemical composition
of carbonfor a carburized 3%Ni-Crsteelwheretwo and the quantity of untransformed austenite may
quenching temperatures are considered. be approximated from the Ms.Together they pro-
The Mf temperature, at which the martensite vide the ability to estimate the retained austenite
reaction ceases, lies approximately 215°C below content from chemical composition. This estima-
the M, (Ref 5). However, the amount of transfor- tion permits the assessment, in terms of retained
mation between M, and Mf is not linear, and austenite, of actual and hypothetical steels when
about 90% of the austenite to martensite trans- subjected to different quenching situations. The
formation takes place during the ftrst 110 °C be- approach has been tested against experimentally
low the Ms. Nevertheless, at high carbon levels obtained and published data, which show that
and assuming that the difference between M, and 90% of the calculated points fall within ±6% of
Mf remains essentially constant, incomplete the measured austenite values and that 80% fall
transformation results when some part of the within ±5% (Fig. 4.6). Figure 4.7 further demon-
transformation range lies below the temperature strates how much the retained austenite might
of the quenchant. The amount of transformation vary within one specifted composition range and
or, alternatively, the volume of untransformed just how significant is the quenching temperature.
An appraisal of the several empirical formulas
for determining M, showed the Steven and
Haynes formula to be the most accurate (Ref 7).
160 Incompletely austenitized steels I
treatedfrom ACt + 100 °C (±30) I

I 500
Base steel com sition, %
<1.1%(Cr+ Mo) / C Mn Ni Cr
I Min 0.12
Max 0.18
3.0 0.60
3.75 1.10



10 200

Fullyaustenitized steels

0.7 0.9 1.1 0

0 0.4 0.8 1.2 1.6
Carbon. %
Carbon, %
Fig. 4.3 Correction curves for use with the Steven and
Haynes formula with author's modifications to original Fig. 4.4 Effect of carbon on the M s temperature. Calcu-
curves shown by dashed lines. When carbon content is lated for the upper and lower extremes of the com-
less than 0.9%, an 830°C soak of over 2 h should pro- posional specification for a carburizing steel at two
duce a fully austenitic structure. Source: Ref 5 quenching temperatures
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
80 / Carburizing: Microstructures and Properties


r-; I I
Vy = e-1.1 0 x 10- 2 (M.-Tq)
20 ~ V
10 ~
<fl- 5
g 0 .....

~ 2 ~

. r-,


0.1 - .• --t
Pureiron-carbon (subzero quench)
0.05 - 0
SAE 52100

•I _1 .1~Io Plai~ carbor
o 50 100 150 200 250 300 350 400 450 500

Fig. 4.5 Austenite content as a function of the difference between

M. and the quenchant temperature. Source: Ref 6

Austenite Layering. In case-hardened surfaces oxidation (Ref 8), or to a reduced matrix carbon
containing retained austenite, the quantity of re- content brought about by the precipitation of car-
tained austenite is often lower at the surface than bides (Ref 9). However, even when there is ap-
at some greater depth beneath the surface. Figure parently neither a reduction of surface carbon
4.8 shows that the calculated austenite contents nor of free carbides at or near the surface, the
agree with those measured away from the sur- surface retained austenite content may still be
face. The lower surface austenite content is gen- lower than at some small distance below the sur-
erally attributed to either changes in the surface face. Plastic deformation of the surface during
chemistry, caused by decarburization or internal

70 Tq=25°C

V 50


<fl- 40

V >
c 30

.tJ0 *'"

:::> Min

, g+ (\l

y rt""'
10 1---f---+~'-7fI'+-

)"'~ O'--_o:::oiiI:;;;..-_...l-_ _'--_--I._ _........___

o o 0.2 0.4 0.6 0.8 1.0
o 10 20 30 40 50 60 70
Retained austenite (measured), % Carbon, %
Fig. 4.6 Comparison of calculated and measured re- Fi~. 4.7 Retained austenite (calculated) in relation to
tained austenite contents for lean-alloy steels (mainly carbon content for a Ni-Cr carburizing steel. See Fig. 4.4
case-hardening grades). for composition.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite / 81

prior machining may be influential, although a stresses have reversed and become compressive,
more likely explanation relates to macrostraining further martensite transformation is inhibited.
during the transformation stage and/or, perhaps, The remaining austenite is protected to some
a degree of austenite stabilization. extent by the presence of compressive residual
Macrostraining can occur during cooling from stresses.
the austenitic condition. During the quench of a The potential contribution of thermal stabiliza-
carburized part, transformation takes place in tion to austenite layering is implied in Fig. 4.1.
the low-carbon core material while the high- The steel has a fixed carbon content and, there-
carbon surface is still austenitic. Therefore, at fore, a fixed M, to M f range, and yet the amount
temperatures approximating M, for the surface of retained austenite varies (e.g., 50 to 60%). The
material, tensile residual stresses develop in major difference between samples producing 50
the still austenitic surface; the greatest amount of and 60% is the rate at which they cooled; a
tension is at the surface. At about the M, of the slower rate, in this instance, gives the higher re-
surface, plastic deformation can take place at tained austenite value.
stresses well below the yield stress of the core This layering aspect of austenite is discussed
material, and the martensite reaction can be because it occurs in case-hardened surfaces,
stimulated. For example, Ankara and West (Ref though perhaps not in every instance; and it
10), using a homogeneous 4%Ni-Cr steel, could have a bearing on both crack initiation and
showed that, with free coaling between the crack propagation under load.
M, and (M, - 2 "C), 7% martensite transformed
from the austenite. When stresses of 70, 140,
and 1035 MPa (10, 20, and 150 ksi) were ap- Austenite in the Microstructure
plied, the amount of martensite produced in-
creased to 10, 20, and 40%, respectively. There-
fore, tensile residual stresses developed at the In microscopic examination of as-quenched
surface of a case-hardened part near the M, carburized surfaces, retained austenite is a white
could lead to more austenite being transformed etching constituent, as are any free carbides, and
than a little deeper in the case where the tensile even martensite is a light etching. Consequently,
stresses are lower. At still lower temperatures be- differentiation of these structural features can
low the Ms, after a fair amount of martensite be difficult without special etchants. Fortu-
transformation occurs and the surface residual nately, tempering, which is applied to most
case-hardened parts, causes fine carbides to pre-
cipitate within the martensite enabling it to etch
more rapidly. Thereby, a much greater contrast
'between the austenite, which remains white etch-
1.0 f---t---+--+-:-+dl---A----4 ing, and the martensite is produced (Fig. 4.9).
Usually the dark etching martensite makes it eas-
ier to see any carbides, especially the network
0.9 t----t----+---*~C1-- carbides.
The white etching austenite volumes are angu-
lar; their shape is determined by the plates of
eft 0.8 ~--+--A~~:y.fIlJ-----l---+
martensite that subdivide each austenite grain.
The size, or coarseness; of austenite volume re-
o 0.7 t---t-lN'~--Itl~f---+----l---4 lates to the prior austenite grain size and the
amount of austenite in the structure, which in
tum are mainly determined by the carbon con-
0.6 f---t---NId-J'------+----l---4 tent, the alloying element content, and the
+. x Observed quenching temperature.
o Calculated values
0.5 L-_---'-U---_-'-_ _...L-_---'_ _-'-
o 10 20 30 40 50
Austenite. % Effect on Material Properties
Fig. 4.8 Relationship between percentages of carbon
and austenite for carburized components where mea-
Influence on Hardness. Retained austenite is
surements of each were made at the surface and at a
depth of 0.375 to 0.5 mm (0.015 to 0.020 in.) from the relatively soft, although it is saturated with car-
surface. Calculated values are also shown. bon. Its coexistence with hard martensite reduces
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
82 / Carburizing: Microstructures and Properties

00 ~

Fig. 4.9 Retained austenite (white) and martensite in the surfaces of carburized and hardened Ni-Cr steel test pieces.
(a) -40% austenite. (b) -15% austenite. Both 550 x

the overall macrohardness of a structure to below 4.11a), the approximate carbon level needed for
that of a structure containing only martensite and austenite to influence the hardness of an
related to the proportions of the two constituents as-quenched steel can be determined. The first
(Fig. 4.10). Figure 4.10 shows carburized, hard- deviation from an essentially straight line (at
ened, and tempered production samples, made -0.35 to 0.4% C) indicates where retained aus-
from nickel-chromium steels with surface carbon tenite begins to affect hardness. For direct
contents of about 0.8%. With a wider coverage quenching, the maximum hardness (800 to 880
of steels and conditions, however, it would not be HV) of the martensite/austenite mix is attained at
unusual to obtain hardness values above the up- between 0.6 and 0.75% C, depending on the steel
per limit of the band as shown. grade. However, the potential hardness is higher
Austenite is retained in small amounts at quite still, though probably at around 0.75% C. At yet
low carbon levels (see Fig. 4.2), and as the car- higher carbon levels, austenite has a marked ef-
bon content increases so too does the retained fect on hardness, especially with direct quench-
austenite content, everything else being equal. ing, as is shown by the sharp decline of hardness
By relating the hardness to carbon content (Fig. (Fig. 4.11).
Influence on Tensile Properties. The room tem-
perature tensile strength and the yield strength in
tension decrease as the retained austenite content
up~rba~d---,3~5% ~I
~ stJls
(cementite network) and direct
increases (Table 4.1, Fig. 4.12). However, high
levels of strain can induce some austenite to
~~ ~ quenched low-alloy steels
martensite transformation, and the amount of
lli600 It: ~~ austenite reduction due to 1% strain was ob-
served to be -7.5 to 10% (Ref 14). Strain in-
~ <,

~5oo duced transformation raises the mean compres-
~ ~ <,
- ...........
sive stress, and the martensite produced is more
~400 r-Lower band---,3.5% Ni
steel (martensite + austenite)
N t..: 0 ductile than that from thermally induced trans-
formation (Ref 13). Conversely, Yen et al. (Ref
<, 15) claim that strain induced martensites in-
crease brittleness, and Franklin et al. (Ref 16)
o 10 20 30 40 50 60 70 80 90 100 provide a reminder that it is untempered, highly
Retained austenite, % strained, and potentially harmful.
Influence on Residual Stresses. The failure of
Fig. 4.10 Influence of retained austenite on the sur-
face hardness of carburized alloy steels. Reheat any austenite to transform during quenching
quenched and tempered at 150 to 185°C means that the volume expansion that should
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite / 83

1000 ,,
, ,,
> 800 .- F:::::'"" !':ingle quench in oil
V <, from 825 °C

600 / -,<,
/ r-,
Direct quench in
from 925°C
Oil" ~

o 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Carbon, wt"k

Direct quench in oil
from 925 °C,
untempered .•, ";';';';';';';' .•..•

600 t---t--#:"--+---+---+---7"<+'~'-:--t-'''o:---1
400 I--~"""'+----I----+---+--~I----''Ir--t*-->:-----l

o 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Carbon, wt"/o
Fig. 4.11 Hardnesslcarbon relationship for untempered martensite in four
case-hardened steels. Source: Ref 11

have accompanied the austenite to martensite stresses are determined by the extent and
reaction did not take place. Therefore, in carbu- sequence of the martensite transformation. Max-
rized and hardened surfaces, the development of imum compression, therefore, occurs at some
residual stresses is in some way related to the distance from the surface where the proportion
amount of austenite not transformed. Koistinen of martensite to austenite is very high, but lower
(Ref 17) made this point and states that both the values of residual compression occur at the sur-
distribution and magnitude of the residual face when the proportion of martensite to aus-

Table 4.1 Tensile test results on through carburized, hardened, and180 O( tempered 4320 steel
Offset yield strength, MPa (ksi)
Ultimate teIBiIe streDgth,
Condition Retained austenite, % 0.02% 0.01% 0.001% MPa(ksl)
4320 (core) 796 (115.5) 523 (75.8) 473 (68.6) 366 (53.0)
Carburized 32.6 604(87.6) 522 (75.6) 456 (66.2) 435 (63.0) 1304 (189.1)
Liquid nitrogen quench 15.0 (a) 1293 (187.5) 1085 (157.4) 577 (83.7) 1503 (218.0)

Carbon content. 1.1%. (a) Strain, <0.002. Source: Ref 12
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
84 / Carburizing: Microstructures and Properties

tenite is low (Fig. 4.13). In Fig. 4.l3(b), the peak specimens with a surface austenite content of
compression coincides with a carbon content of 80%, and Brugger (Ref 23) observed that, when
-0.55%. However, Salonen (Ref 18) determined the austenite contents of a case-hardened
a value of 0.6% C at peak compression, whereas 15Cr-16Ni steel were 40 and 20%, the bending
others (Ref 19) agreed with Koistinen. fatigue limits were respectively 16 and 4% below
The magnitude of the residual stresses and the the essentially austenite-free condition. Pacheco
residual stress distribution are influenced by
chemical composition (including carbon content)
and quenching method (including cooling rate). Depth below surface, mm
In Fig. 4.14, two steels are compared. One steel o 05 10 15
30 I 10
has a martensite/austenite outer case, and the SAE8620
other steel also contains some bainite. 25 - slight surface 0 o
Influence on Fatigue Resistance. Fatigue crack <P-
initiation and early propagation at and in a >'20
case-hardened surface are strongly influenced by c:
CD \ ,
the inherent strength of the material and the pre- ~ 15 ,
I\. ,,
vailing residual stresses. High values of com- "C
~ 10 I tres

~ ........., v
pressive residual stresses are favored to negate
applied tensile stresses. The presence of retained ! 5 ~
austenite, however, reduces both the strength (as
implied by hardness) and the compressive resid- o -50
o 10 20 30 40 50 60
ual stresses. Therefore, it would be expected to
Depth below surface. 10-3 in.
lower the fatigue resistance to a degree depend-
ent on the relative proportions of martensite and
retained austenite. This tendency is established Depth belowsurface, mm
by Wiegand and Tolasch (Ref 21), who state that 0 0.5 1.0 1.5

the bending fatigue limit of unnotched

case-hardened test pieces decreases as hardness
1.2 60
1 SAEll18
10 c.
'/ij 100
,, " ~ ~
falls below about 680 HV. With notched sam- gO.8 0 ui
lJ) 0 ui
~ ~
ples, the fatigue limit progressively falls as the c:
_10 1il 1il
8>0.6 -100 'il
hardness decreases (Fig. 4.15). In terms of actual .e
::J ::J
austenite contents, Razim (Ref 22) obtained a 5e 0.4 -20'ijj
-200 a:
25% reduction of bending fatigue strength from
~0.2 -30
0 0 -40
90 0 10 20 30 40 50 60
600 Depthbelowsurface, 10-3in.

\ (b)

g,j 70
\ o
Depth below surface, mm
0.5 1.0 1.5
\\ ~

SAE 514h

10 ~

o Stress ,,'
I \ >. 40 , I-- o ~
50 .~ gf o
40 I - A4098AQ&T \ *'"

::J 30
'~ "
, " ,
-100 'iii

04012600 & T
30 f- • 40126 00 +
liquid nitrogen & T
\ 150



~ --'
.; , J -20 ~


III -200

o 10 20 30 40 50 60 o 10 20 30 40 50 60
Retained austenite. % Depth below surface, 10-3 in.
Fig. 4.12 Dependence of stress for first detectable (e)
plastic strain (-0.0001) on retained austenite content.
QQ. oil quenched; T, tempered; AQ, air quenched. Fig. 4.13 Retained austenite and residual stress distri-
Source: Ref 13 butions in case-hardened test pieces. Source: Ref 17
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite / 85

Distance from surface, mm Distance from surface, mm

o 0.2 0.4 0.6 08 1.0 12 o 02 04 06 08 10 12 14
o o
50 ~RC
-100 -50
[....--- I - -

gf -100
V ~
iii /
V -
~ -150
/ 'C

!\ J
SAE 1526

SAE4140 V
-500 I I -250 I I

Bainite 0 0 5 45 -100 Bainite 32 22 20 47 80 90 100

Martensite 88 78 80 45 0 Martensite 60 70 72 47 10 20 0
Austenite 12 22 15 10 0 Austenite 8 8 8 6 0 0 0

Microconstituent content, vol% Microconstituent content, vol%

(a) (b)

Fig. 4.14 Residual stress distributions in two oil-hardened carburized gears. Source: Ref 20

Bml Through-hardeningand tool steels

11ll1l Throu~h hardened steels-notched 1250
120 Case- ardened steels-notched
&iii Case-hardening steels
'" 1000 Ol~
£0, ~ :2.
~ ~~ 80 .,£
.c£ 750
g>g> g>~
~~ ~(;)
e: e: .,
~ Ql
III 500 ~ ~


0 0
100 300 500 700 900 1100
Hardness, HV5

Testpiece diameter 12 mm
or Constant torque 1000 III
OlE OlD.
.£: :2
e:~ 80 750 .,-
.~~ ~1i)
1U~ e: .,
500 ~ ~

~~ 40 «$
~ 250

0 0
0 1000
Hardness, HV5
Fig. 4.15 Rotating beam fatigue strength of case-hardening, through- hard-
ening, and tool steels as a function of surface hardness. Source: Ref 21
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
86 / Carburizing: Microstructures and Properties

1800 ~----'T---r---.....-----, 261 .,

• Plasmacarburizing i
• Gas carburizing '5105 440 MPa
(45 kglmm2)
1600 232 lii

~ 0 0 605 MPa (62 kglmm2) o

8!. 'iii c
o 0
:::ii 1400 203 "":
.~ 104
l ~Q)
Q) :::l :::l
:::l .2l ..g»
~ 1ii S
If 1200 174 u;
.§ 103
0 10 20 30 40 50 60 70
1000 ASTM grain size 145 Retained austenite. %
Fig. 4.17 Effect of retainedausteniteon the impact-fatigue
resistance of a 1.45C-ll.5Cr steel.Source:Ref 26

20 30 40
ing austenite content. At intermediate stress lev-
Retainedaustenite. %
els, the austenite content has no effect. At low
Fig. 4.16 Fatigue limits of plasma and gas-carburized values of stress, an increased austenite lowers the
specimens as a function of retained austenite content. fatigue resistance (Fig. 4.17).
The results of Panhans and Foumelle (Ref 27)
do not altogether agree with the findings of
and Krauss (Ref 24) confirmed that not only Kozyrev and Toporov. Whereas Panhans and
should the austenite be minimal but it should be Foumelle (Ref 27) did find austenite to be bene-
tine and evenly distributed (Fig. 4.16). It is, ficial at less than 1 x 1()4 cycles, they also found
therefore, generally accepted that retained aus- it to be beneficial at more than 1 x 1()6 cycles
tenite is detrimental to the bending fatigue limit and marginally inferior between the two (Fig.
(low-stress, high-cycle fatigue) and probably 4.18). This contradiction is difficult to explain on
also to the torsional fatigue limit. the basis of either plastic deformation of the aus-
At high applied stress levels (high in the finite tenite or the formation of strain-induced
life part of the S-N curve), on the other hand, it is martensite. Brugger (Ref 23) did not find austen-
possible that retained austenite is beneficial by
slowing down the crack growth rate. The rate of 1200 , . - - - - . . , - - - - , - - - - r - - - . . ,
crack growth slows when the strain ahead of a
AISI E 9310 carburized
crack tip (propagating in an austenite containing and quenchedfrom 843°C
surface) induces the austenite to martensite reac- in oil at 60°C
tion, thereby raising the strength and increasing 1100
the compressive residual stresses (Ref 25). I'll
At relatively low stresses close to the fatigue :::ii
limit, there is insufficient strain to induce the "0
austenite to martensite transformation. It is only a. 1000
high applied cyclic stresses that bring about that .,
transformation, which equates to fatigue lives of l!!
less than about 1 x 1()4 load cycles. These points
appear to be confirmed by Kozyrev and Toporov
(Ref 26), who isolated the effect of austenite on
the impact fatigue resistance using a high-carbon
12% Cr alloy steel. The quenching temperature 800 l..-_ _..l-_ _--l-_ _- - - J L . . - _ - - =
was varied to produce different austenite con- 103 104
tents without influencing the grain size. They es- Cycles to failure
tablished that at high levels of applied stress the Fig. 4.18 5-N curves for case-hardened 9310 steel
impact fatigue resistance increases with increas- (untemperedl. Source: Ref 27
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite / 87

ite to be beneficial at any stress level, nor did tically deform under rolling contact pressures. In
Pacheco and Krauss (Ref 24). rolling contact disc tests, normally the more
To confuse the issue a little more, Szpunar and highly loaded surfaces have longer lives than
Bielanik (Ref 8), studying the crack propagation those tested at intermediate load levels. This re-
characteristics of case-hardened steels, report sult means that the plotted test data have a ''C''
that the two steels studied had different crack shape rather than the more familiar S-N plot
propagation behaviors. Steel 20HNMh (SAE (Fig. 4.20). The effect has been observed in
8620) exhibited a maximum crack propagation case-hardened steels with and without retained
rate when the retained austenite content was austenite and in high-strength nitrided steels with
23%, after which the crack propagation rate de- no retained austenite. Having said that, an aus-
clined (Fig. 4.19a). Figure 4.19(a) indicates that tenite containing case-hardened surface deforms
over 40% retained austenite appeared to be effec- more readily than a wholly martensitic surface or
tive in dulling the crack or even preventing its a martensite/bainite surface, thereby contributing
growth. The 18HGT (Mn-lCr-Ti) steel, on the
more toward improved durability at higher stress
other hand, showed little change of the crack
propagation rate for retained austenite contents
Regarding the influence of contact stressing on
up to about 30% (Fig. 4.19b). Above about 30%,
surface hardness, case-hardened surfaces with up
the behavior depended on which of the three load
to about 20% retained austenite and tempered at
amplitudes was used. At a higher load amplitude,
the crack propagation rate increased with in- less than 150°C harden by rolling contact in a
creasing austenite content. Whereas at a lower range of 85 to 120 HV, whereas those surfaces
load amplitude, the crack growth rate decreased tempered at 150 to 250°C harden by only -35
as the austenite increased with no growth when HV. Razim (Ref 28) noted that in surfaces con-
the austenite content was about 80%. Therefore, taining -50% austenite hardness increases, in
the energy absorbed to induce the austenite to general, from 500 to 1000 HV due to rolling,
martensite reaction could not be available for whereas surfaces containing no austenite are
crack propagation. hardly, if at all, affected. The increased hardness
Influence on Contact Fatigue. In bending fa- in this case is primarily the result of working;
tigue situations, any benefits of retained austen- both slip lines and induced carbide precipitation
ite appear because of the austenite-martensite re- were observed. These data provide a rule of
action. However, in rolling contact situations, the thumb measurement of the effect of austenite
cold working property of the material is more content (in case-hardened surfaces) on the hard-
likely to be important, that is, its ability to plas- ness increases caused by fairly heavy rolling

24 24
Akrange Akrange
o 18-24 MPa {in _ 018-24 MPa-fm
20 20 - •
\ o 20-26 MPa {in 020-26 MPa-fm V
CIl 16 / • 21-28 MPa {in
CIl 16
• 21-28 MPa-fm
~E 1 I
i 12
h 12
I r\\ - V



4 ~ r- 4

20 40 60 80 100 20 40 60 80 100
Retainedaustenite, vol"lo Retainedaustenite, vol"lo
(a) (b)
Fig. 4.19 Average propagation rate of fatigue cracks in carburized cases of (a) 20 HNMh and (b) 18 HGT
steel depending on percentage of retained austenite and on load amplitude. Source: Ref 8
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
88 / Carburizing: Microstructures and Properties

contact (i.e., for each 20% austenite, the hard- containing surfaces are superior to those contain-
ness increases by about 130 HV). ing only small amounts. Diament also showed
The contact fatigue tests carried out by Razim that as the contact fatigue tests progress the re-
(Ref 28) showed that the surface fatigue resis- tained austenite content within the case was al-
tance increased as the retained austenite content tered and the residual stress distribution was
increased (up to about 55%). Further, in compar- modified.
ison with other microstructures likely to be de- Vinokur (Ref 31, 32) examined the effect of
veloped during case hardening (e.g., carbides), quenching a 0.96% C alloy steel from tempera-
those containing austenite gave the most favor- tures within the range 730 to 930°C (in 20°C
able results (Fig. 4.21) at all stress levels. The increments) and found that as the quenching
whole of the fatigue curve was raised as austen- temperature increases, the amount of carbide in
ite increases. Balter and Throvskii (Ref 29) and the final microstructure decreases, and the re-
Diament et al. (Ref 30) agreed that high austenite tained austenite content increases. The best con-
tact fatigue resistance is achieved by quenching
from the ACm (810 °C for the steel in test), fol-
40.000 r - - - - - - : : - 1 - - - - - - - - , lowed by tempering at 150°C for 2 h (Fig. 4.22).
The maximum resistance to contact fatigue re-
lates not only to the austenite content but also to
35.000 1----F.--7,~----l-------l
the amount of martensite (65 to 70%) and the
cl near absence of carbides in the structure (matrix
ui 30.000 ~----=~-~~~:---___1 carbon 0.8%). The retained austenite was 32%
£ before testing and -24% after testing, and the
optimum structure for fatigue resistance more or
25.000 t----------i"'-:::--,,------j less coincided with the initial peak hardness.
• Tempered at loo·C The surface residual stresses developed during
A Tempered at 150·C

20,000 ' - - - - - - - - - - - - ' - - - - - - - - '

case hardening can play a part in the contact fa-
106 tigue life of a component. Barczy and Takacs
Loadcycles (Ref 33), after quenching carburized planet pins
Fig. 4.20 Rolling contact fatigue plots for carburized into a hot quenchant, found a sharp tensile peak
and hardened 3Ni-Cr steel discs. SH = (Ibfin. of face residual stress 0.2 mm beneath the surface, and
width}f{relative radius of curvature) this near-surface stress peak was responsible
for surface spalling after a very short life. By
raising the surface carbon content and the aus-
400 tenite content (15%), the tensile stress peak

p/ No carbides


/' networks

10 hf-t--t--t-~t:::::::l4 4 .e
0 0 8
730 no 810 850 890 930
o 10 20 30 4 50 60 Austenitizing temperature. °C
Retained austenite, vol% Fig. 4.22 Effectof austenitizing temperature on
Fi~. 4.21 Effect of austenite on the pitting resistance of hardness, amount of retained austenite, and con-
caiburized gears. Taken at 10 7 cycles, which approxi- tact fatigue strength of 90KhGNMFL steel under a
mates to the "knee" of the S-N curve. Source: Ref 28 3.43 GPa (350 kgffmm 2 ) load. Source: Ref 32
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite /89

Table 4.2 Effectof quenching temperature on bend and impact strength of case·hardened 15Cr-6Ni steel
QueD<h temperature, ·C Bend strength, MPa Impact strength, MPa Impact fatigue strength, MPa
900 40 57 1500 2500 670
860 20 59 1420 2500 770
820 o 61 1390 2500 790
790 o 61 1250 2800 800

Source: Ref 23

shifted to 0.4 mrn below the surface, and the ser- austenite (and nickel content) increases, the ini-
vice life improved. Further increase of the car- tial crack strength also increases (Fig. 4.24). The
bon content raised the austenite content to 30% bend and impact test results for direct quenched
and flattened the residual stress distribution to specimens are shown in Fig. 4.25. The results for
virtually zero. The outcome was that the service the double quenched condition (with a finer
life was appreciably improved. In this instance, grain size) generally produced even better val-
compressive residual stresses resulting from heat ues, particularly when the nickel content ex-
treatment did not help, and the adverse effect ceeded about 2 to 3%. Tempering was especially
of the sharp tensile peak dominated events. beneficial with respect to the bending strength.
Compressive residual stresses, whether caused In terms of fracture toughness tests (Ref 37),
by heat treatment or plastic deformation and the KIc value, which decreases with increasing
strain-induced martensite, are thought to be ben- carbon content, tends to increase with nickel
eficial because they can "squeeze" a crack, content and with retained austenite content (Fig.
which increases the crack-face friction that, in 4.26). Here, the critical crack size increases with
tum, slows down the crack propagation rate the nickel content. Consequently, a steel with
(Ref 34). more than 3% Ni is regarded as having a high
It seems, therefore, that retained austenite is initial crack resistance, in keeping with Ref 36.
beneficial under rolling and rolling with sliding Influence on Wear Resistance. For straightfor-
conditions, assuming the contacting surfaces re- ward abrasive wear situations, a high surface
main reasonably separated throughout the work- hardness is the main property requirement. This
ing temperature range by adequate lubrication. surface hardness entails developing a surface
Influence on Bending and Impact Fracture microstructure of high-carbon martensite with,
Strength. Whereas Brugger's fatigue tests (Ref perhaps, some spheroidized carbides and mini-
23) found no merit in having retained austenite mum retained austenite. The influence of re-
in the carburized case, the accompanying bend tained austenite on the adhesive wear of
tests showed that samples with the highest aus- case-hardened surfaces is complicated by the re-
tenite content produced better results (Table 4.2).
This result might be a reflection on the ability of
austenite to yield at high surface stresses caused 3500
by bending. In these tests, however, the different
austenite contents were achieved by using differ-
ent quenching temperatures. As a consequence,
the results might be influenced more by the con-
dition of the material at the core rather than by ~ 2500

the presence of austenite at the surface. The im-

pact strength did not seem to be affected by the c: 2000

presence of surface retained austenite, which is ~
in agreement with other works (Ref 19). :-s 1500

Razim considered that case toughness is indi- ~ - 20NiCrMo6
cated by the initial crack strength of notched test o •••• 15CrNi6
1000 I - ....... 16MnCr5
pieces (Ref 35) and showed that as the retained - 20MoCr4
austenite (and carbon content) increased, the ini- 500 I I
tial crack strength decreased (Fig. 4.23). On the o 20 40 60 80 100
other hand, Thoden and Grosch (Ref 36), work- Retained austenite, %
ing with samples carburized to a surface carbon Fig. 4.23 Effect of increasing retained austenite on
content of 0.65%, showed that as the retained crack initiation strength. Source: Ref 35
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
90 / Carburizing: Microstructures and Properties

25 25

16Ni7 16Ni18

Heat treatment:
20 computer controlled carburizing 20
925°C/oil + 840°C/oil + 180 °C for 1h
Eht= 0.6 mm

15 z 15
-0 "'-0"
Ol Ol
a. a.
§ 10 § 10

51-+--H++--,-----+-----1 5H----+-----\l'-\--,it-+-----1

250 500 750 o 250 500 750
Time, I!S Time, its
Fig. 4.24 load-time curves for carburized DVM/DIN 50115 specimens. The initial crack occurs at maximum
load. Heattrealment: computer controlled carburizing at925 °C in oil + 840°C in oi I + 180°C for 1 h, Eht = 0.6 mm.
Source: Ref 36

spective instabilities of austenite and martensite erated by plastic deformation and friction en-
in the microstructure. courages carbide precipitation (primarily from
Under contact loading, which may involve within the martensite) and softening, which can
both rolling and sliding, austenite can be plastic- reduce wear resistance. Heat generated at a
ally deformed and strengthened, resulting in a working surface can also impair the efficiency of
greater resistance to wear. Conversely, heat gen- the lubricant, favoring the adhesive wear process.
Surface roughness is regarded as detrimental
to the efficiency of the lubricant, because surface
asperities can penetrate the lubricant film and
3000 Direct quench 925°C in oil

o • Eht = 0.6 mm
o • Eht = 1.0 mm
t l Tempered 25 Heat treatment computer controlled carburizing, 925 'C inoil
L As-quenched
•• Eht= 0.6 mm
Eht = 1.0 mm
D •

~ 2000tiE:=:l:=~~---1-:Q:=:J_-o-l



-- »>
~ --
6> J;
c: 16Ni7
'6 16MnCrs 16Ni14 16Ni18
c: 10

4 5 o 2 3 4 5

Nickelcontent, wt%
(8) (b)

Fig. 4.25 Bending (a) and impact (b) strengths of four carburized and direct quenched steels. Source: Ref 36
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬

Table 4.3 Effect of retainedausteniteon the scuffing tendencyof steel

Steel Retained austenite, % Ratl!ll!(a) AGMA2001-C95welding factor, Xw
Stainless steel 100 0.32 0.45
Case-hardenednickel steel >20 0.80 0.85 (>30% 1r )
Through-hardenedsteel and -20 1.00 1.00 (20-30% 1r )
case-hardenedchromium steel
Case-hardenednickel or chromium steel <20 1.20 1.15

(a) The higher the rating is, the greater the scuffingresistance will be. Source: Ref 38

make contact with similar asperities on the mat- tween scuffing resistance and retained austenite
ing surface. The removal of surface asperities content.
caused by wear or deformation (probably as- Grew and Cameron (Ref 41) suggest that aus-
sisted by the presence of retained austenite), tenite has a lower affinity than martensite for the
therefore, is considered to be a favorable if not surface-active compounds contained in lubricat-
crucial happening. It redistributes the load and ing oils; therefore, it will be more difficult to lu-
reduces the frictional effects, provided reason- bricate. In their tests, which relate the coefficient
able lubrication is maintained. Further, with the of friction to the frictional temperature, a carbu-
surface asperities removed, the local contact rized and hardened 4%Ni-Cr-Mo steel with 5%
pressure peaks within the mating surfaces are retained austenite remained fairly stable,
also removed as well as the risk of microflaking. whereas with 25% austenite some instability oc-
There are instances, however, where highly pol- curred at 150°C. Following a 180°C temper, the
ished surfaces can be difficult to lubricate, and a instability of the surface originally containing
loss of lubricant in such circumstances could 25% austenite did not occur until a 180°C fric-
lead to adhesive wear. tion temperature was reached (Fig. 4.28). The
In their approach to predicting the scuffing observed instabilities relate to the addition of
tendencies of gears, Niemann and Seitzinger surface active compounds, and yet, coincidence
(Ref 38) introduce an Xw factor to account for or not, the observed instabilities seem to relate to
the potential adverse influence of austenite the already known thermal instability of austen-
(Table 4.3). However, the low Xw factor given ite. The second stage of tempering commences at
to stainless steel and attributed to 100% aus- about 150°C, and for an already tempered sur-
tenite may, in fact, be caused by the nickel con- face, the second stage of tempering recom-
tent and the absence of carbide precipitates, as mences at temperatures above the original tem-
shown in Fig. 4.27 (Ref 39). Roberts (Ref 40) pering temperature. As a reminder, the second
points out that both the nickel and the chro- stage of tempering involves the transformation of
mium austenitic stainless steels are notoriously retained austenite to bainite (ferrite with carbide
difficult to lubricate. Roberts (Ref 40) goes on precipitates).
to assert that there is no direct relationship be-


.z:20I__----+-------'==-......,~--I__----"'--..L_'I (90)

o -3V:P1O Ni
CJ-1V2% Ni OL-_ _L -_ _L -_ _L -_ _ .&:-_~

e Low Ni o 4 10
10 '--_ _--'- ....1- ' - -_ _-1
Alloying element content, at.%
0.6 0.7 0.8 0.9 1.0
Fig. 4.27 Effect of alloy content in Fe-' %C materials
Carbon, % on the critical temperature of a mineral oil. Data in paren-
Fig. 4.26 Effect of carbon and nickel contents on frac- theses indicate percentage retained austenite content.
ture toughness. Source: Ref 37 Source: Ref 39
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
92 / Carburizing: Microstructures and Properties

Laboratory tests to determine the effect of creases (Fig. 4.30). It was observed that when
austenite on scoring resistance of case-hardened there are small quantities of retained austenite
surfaces give conflicting information. For ex- in the microstructure, initial seizure occurs by
ample, Kozlovskii et al., using a low- to me- shearing at an austenite volume and spreads
dium-speed roller test, concluded that a high re- from there as the load increases. When the sur-
tained austenite content increases the scoring face has a high retained austenite content, the
resistance by virtue of the high capacity of aus- surface plastically deforms to spread the load;
tenite to work harden. On the other hand, hard destructive seizure, when it occurs, then takes
low-ductility surfaces containing only small place across the whole contact area, that is,
amounts of retained austenite are more likely to there is no initial seizure point.
suffer scoring damage at lower pressures (Fig. Retained austenite and its contribution to the
4.29) (Ref 42). Supporting that view to some adhesive wear process could relate to how the
extent, Manevskii and Sokolov produced results mating surfaces are "run-in" and to the response
from case-hardened and carbonitrided test of the material to the running-in process. For
pieces suggesting that the hardness for best sei- example, experience with aluminum bronzes
zure resistance is approximately 580 to 600 HV; for worm-wheel applications indicated that a
harder surfaces tended to score at lower pres- material initially in a soft condition work hard-
sures (Ref 43). Terauchi and Takehara stated ened during roll/slide tests to a final hardness
that with surfaces of less than 500 HV, the sur- level without any signs of scuffing. For such a
face hardness increases with repeated rolling, condition, the load carrying capacity was good.
whereas with martensite- only microstructures, On the other hand, when material was made
the hardness diminishes during testing if it with an initial hardness equal to the final hard-
changes at all (Ref 44). They concluded, how- ness of the softer alloy, scuffing readily oc-
ever, that martensitic surfaces have the highest curred without their being much in the way of
scoring resistance, and that the scoring resis- work hardening. If the same trend applies to
tance decreases as the austenite content in- the scuffing behavior of case-hardened sur-

0.8 0.8
Case hardened
:::L Bainitic microstructure :::L Martensite + 25% retained austenite
c: 0.6 c: 0.6
15 0.4
"0 0.4
i: i:
Q) Q)
:2 +-
0.2 ~ 0.2

80 120 160 200 80 120 160 200

Temperature, °C Temperature,oC

0.8 0.8
Case hardened Case hardened and tempered (175-185 0C)
:::L :::L
Martensite + 5% retained austenite_
c: 0.6 I--- c: 0.6 _ Martensite +25% retained austenite

~ U
0.4 (5
~ 0.2 ~ 0.2

80 120 160 200 80 120 160 200

Temperature, °C Temperature, °C
Fig. 4.28 Friction temperature curves from Bowden-Leben machine. Tungsten-carbide slider onto uncarburized,
case-hardened, and case-hardened and tempered 4%Ni-Cr-Mo test pieces. Source: Ref 41
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite / 93

faces, then just sufficient retained austenite to ting, the austenite content should be kept to a
assist the running-in process would seem to be minimum (preferably without refrigeration). If
in order. The question then is, what is "just suf- the converse applies, then a retained austenite
ficient"? content approaching 35% is more appropriate. If
occasional tooth bending overloads are expected,
then to deter initial cracking it might be prudent
to aim for an austenite content of about 25%. If a
Control of Retained Austenite life of less than I x I Q4 cycles at high loads is re-
quired, then much higher austenite contents can
The retained austenite content at the surface of be considered. If, on the other hand, adhesive
a carburized steel is influenced by the alloy con- wear (scoring, scuffing) is the more likely failure
tent of the steel, the surface carbon content, and mode, then it is necessary to lower the austenite
the quenching temperature, all of which deter- content.
mine the Ms. The temperature of the quenchant Heat Treating. It should be apparent that no
also contributes to the as-quenched retained aus- single value for retained austenite content will
tenite content. Primary control, therefore, must satisfy all requirements of a gear tooth. The
be by manipulation of these variables. foregoing discussion is very nice in theory, but
Considerations Regarding Properties. Whereas in practical terms it is not altogether realistic. If
certain standards quote maximum retained aus- a designer had the time to determine the best
tenite contents, there is some latitude for the austenite content for a particular gear, could the
manufacturer to adjust the amount within those heat treater comply? In commercial heat treat-
limits (i.e., 0 to 35% American Gear Manufac- ment, the best an operator can do, given a steel
turers Association). Before deciding on the sur- grade and quenching method, is adjust the sur-
face carbon content and the quenching tempera- face carbon content to obtain a retained austen-
ture for a chosen steel grade, it is necessary to ite content within the recommendation of the
establish what level of retained austenite can be standard.
tolerated. Making these decisions requires some Post case-hardening refrigeration is an effec-
knowledge of how the component functions and tive means of reducing retained austenite, but it
how it will be loaded in service. In dealing with can have drawbacks in its influence on material
gears, each type of loading, or action, must be properties. Therefore, carburizing and hardening
considered. For example, if the safety factor for processes should avoid the formation of exces-
tooth bending is small compared to surface pit- sive amounts of austenite in the first place. If a

2500 ~-----'
1100 M1,10.7% RA
CJ Austenite at surface M2, 18.1%RA
1000 c- CJ Austenite at 0.15 mm _ M3, 49.2% RA

-- f---

~ 800 - f-
.>< ~ 1500 -
700 - - co
!!l Gl
'" :; r--
.~ 600 - - N
a; 1000
'8 500 - -
400 f--- -

300 - - ,...
f--- -
o 10 20 30 40 50 60 70 M1 M2 M3
Retained austenite, % Specimen

Fig. 4.29 Effect of retained austenite on scoring re- Fig. 4.30 Relation between seizure load and percent-
sistance. Source: Ref 42 age of retained austenite (RA). Source: Ref 44
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
94 / Carburizing: Microstructures and Properties

subzero treatment is unavoidable, then a shallow benefit contact fatigue. Conflicting data exist
rather than a deep freeze should be considered, regarding adhesive wear. In all cases, the aus-
as too should both prior and post tempering op- tenite and accompanying martensite should
erations. The subject of refrigeration is covered be fine and evenly distributed.
in more depth in Chapter 7, "Postcarburizing • Standards: For the lower quality grade neither
Thermal Treatments." ANSI!AGMA nor ISO provides a specifica-
Surface Working. The properties of surfaces tion. For the middle and the highest grades,
containing retained austenite can be favorably ANSI!AGMA quotes 30% maximum retained
modified by mechanical methods that induce austenite, whereas ISO calls for 25% maxi-
plastic deformation, such as shot peening or mum.
surface rolling. Retained austenite, being rel-
atively soft, is work hardened by these pro-
cesses with the added benefit that the surface REFERENCES
residual stresses are made more compressive.
It is important, however, that the balancing I. A Rose and H.P. Hougardy, Transformation
tensile residual stresses do not peak just be- Characteristics and Hardenability of Carbu-
neath the worked layer. Correctly executed rising Steels, Transformation and Harden-
surface working is capable of not only over- ability in Steels, Climax Molybdenum Co. of
coming any adverse effects 'of retained aus- Michigan, 1967, p 155-167
tenite but also of raising the fatigue strength 2. E.C. Rollason, "Fundamental Aspects of Mo-
of the part to well above that obtained solely lybdenum on Transformationof Steel,"Climax
by thermal means. Molybdenum Co. of Europe Ltd.
3. R Priestner and S.G. Glover, Stabilisation
Effect in a High Carbon Nickel Steel, Physi-
cal Properties of Martensite and Bainite,
Summary Special report 93, The Iron and Steel Insti-
tute, 1965, p 38-42
Retained austenite is more or less unavoidable 4. W.S. Owen, Theory of Heat Treatment, Heat
in quenched carburized steels. The amount of re- Treatment ofMetals, Hatte Books Ltd., 1963,
tained austenite is determined by the carbon con- P 1-28
tent, quenching, and quenchant temperatures. 5. AG. Haynes, Interrelation of Isothermal and
For most applications some retained austenite is Continuous-Cooling Heat Treatments of
acceptable; if in excess, it can lead to grinding Low-Alloy Steels and Their Practical Signifi-
problems. cance, Heat Treatment of Metals, Special re-
port 95, The Iron and Steel Institute, 1966, p
• Preprocess considerations: It is normal to se- 13-23
lect steel for a component based on size and 6. D.P. Koistinen and RE. Marburger, A Gen-
eventual duty. Therefore, the possibility of eral Equation Prescribing the Extent of the
producing high retained austenite contents at Austenite-Martensite Transformation in Pure
the surface can be anticipated. Iron-Carbon Alloys and Plain Carbon Steels,
• In-process considerations: Carbon-potential Acta Metall., Vol 7, 1959, p 59-60
control and quenching methods are means of 7. c.Y. Kung and J.J. Rayment, An Examina-
controlling the austenite content. Generally, tion of the Validity of Existing Empirical
lean grades of steel are direct quenched, and Formulae for the Calculation of M, Tem-
the more highly alloyed grades are reheat perature, Metall. Trans. A, Vol 13, Feb 1982,
quenched. p 328-331
• Postprocess considerations: If retained austen- 8. E. Szpunar and 1. Beilanik, Influence of Re-
ite is unacceptably high, consider requenching tained Austenite on Propagation of Fatigue
from a lower temperature (watch distortion as- Cracks in Carburised Cases of Toothed Ele-
pect). Alternatively, use shallow refrigeration if ments, Heat Treatment '84, The Metals Soci-
the adverse effects of the process can be toler- ety, p 39.1-39.9
ated. 9. L.A Vasil'ev, Retained Austenite and Hard-
• Effect on properties: Retained austenite re- ness of the Carburised Case on Steel
duces hardness, abrasive wear resistance, and 18Kh2N4VA after Quenching in 'IWo Differ-
bending-fatigue strength, but it is thought to ent Media, Met. Sci. Heat Treat., Vol 14
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
RetainedAustenite/ 95

(No.2), Feb 1972,p 58-59 22. C. Razim, Influenceof Residual Austeniteon

10.O.A. Ankaraand DRE West, Investigation of the StrengthProperties of Case-Hardened Test
Transformation Plasticity during Martensite Pieces during Fatiguing, Harterei-Technische
Formation in a Medium Alloy Steel, Physical Mitteilungen (BISI 6448), Vol23, April 1968,
Properties of Martensite and Bainite, Special P 1-8
report 93, The Iron and Steel Institute, 1965, 23. H. Brugger, Werkstoff und Wiirmebehandlung-
p 183-192 seinflusse auf die Zahnfujitragfahigkeit,
11.K. Bungardt, E. Kunze, and H. Brandis, VDI-Berichte, (No. 195), 1973, P 135-144
Betrachtungen zur Direkthartung von 24. lL. Pacheco and G. Krauss, Microstructure
Einsatzstahlen, DEW-Technische Berichte, and High Bending Fatigue Strengthof Carbu-
Vol5 (No.1), 1965, P 1-12 rised Steel, J Heat Treat., Vol7 (No.2), 1989,
12.RW. Neu and S. Huseyin, Low Temperature p77-86
Creep of a Carburised Steel, Metall. Trans. A, 25. M.A. Zaconne, lB. Kelley, and G. Krauss,
Vol 23, Sept 1992,p 2619-2624 Strain Hardening and Fatigue of Simulated
13.RH. Richman and RW. Landgraf, Some Ef- CaseMicrostructures in Carburised Steel, Corf.
fects of RetainedAusteniteon the FatigueRe- Proc. Processing and Performance (Lakewood,
sistanceof Carburized Steel, Metall. Trans. A, CO), ASM International, July 1989
Vol 6, May 1975,p 955-964 26. G.Y. Kozyrev and G.V. Toporov, Effectof Re-
14.R.I. Johnson, 'The Role of Nickel in Carbu- tained Austenite on the Impact-Fatigue
rising Steels," Publication A 1205, Interna- Strength of Steel, Metal Sci. Heat Treat., Vol
tional NickelCompany, Inc. 15 (No. 12),Dec 1973,P 1064-1066
15.X. Yen, D. Zhu, and D. Shi,The StressInduced 27. M.A.Panhansand RA Fournelle, HighCycle
Phase Transformation of Carburising and FatigueResistance of AISI E 9310 Carburised
Carbonitrided Layers and Their Contact Fa- Steel with Two Different Levels of Retained
Austenite and SurfaceResidualStress, J Heat
tigue Behaviour, Heat Treat. Met. (China), Vol
Treat., Vol 2 (No.1), 1981,P 54-61
1, 1984,P 31-37
28. C. Razim, Effects of Residual Austenite and
16.l Franklin, P.Hill, and C. Allen,The Effectof
Recticular Carbideson the Tendency to Pitting
Heat Treatmenton RetainedAustenite in a 1%
of Case Hardened Steels, thesis, Techn.
Carbon-Chrornium Bearing Steel, Heat Treat. Hochschule Stuttgart, 1967
Met; Vol 2, 1979,p 46-50 29. M.A. Balterand M.L. Throvskii, Resistance of
17.D.P. Koistinen, The Distribution of Residual Case-Hardened Steel to Contact Fatigue,
Stressesin Carburised Cases and Their Origin, Met. Sci. Heat Treat., (No.3), March 1966,
Trans. ASM, Vol50, 1958,P 227-241 p 177-180
18. L. Salonen,The Residual Stressesin the Car- 30. A Diament, R El Haik, R Lafont, and R
burisedLayersin the Caseof an Unalloyed and Wyss, "Surface Fatigue Behaviour of the
a Mo-Cr Alloyed Case-Hardened Steel after Carbonitrided and Case-Hardened Layers in
Various Heat Treatments, Acta Polytech. Relation to the Distribution of the Residual
Scand., Vol 109, 1972,P 7-26 Stresses and the Modifications of the Crystal
19.D.E. Diesburg, C. Kim, and W. Fairhurst, Lattice Occurring during Fatigue" (BISI
Microstructural and ResidualStressEffectson 12455), paper presented at 25th Colloque Int.
the Fracture of Case-Hardened Steels, paper (Caen, France), 29-31 May 1974, Interna-
23, Heat Treatment '81 (University of Aston), tional Federation for the Heat Treatment of
The Metals Society, Sept 1981 Materials, 1974
20. M.M. Shea, Residual Stress and Micro- 31. B.B. Vinokur, S.E. Kondratyuk, L.I.
structurein Quenched and Tempered and Hot Markovskaya, RA Khrunik, AA Gurmaza,
Oil Quenched Carburised Gears, J. Heat and Y.B. Vainerman, Effect of Retained Aus-
Treat., Vol 1 (No.4), Dec 1980, P 29-36 teniteon the Contact Fatigue Strengthof Car-
21. H. Wiegand and G. Tolasch, The Combined burised Steel, Met. Sci. Heat Treat., (No. 11),
Effect of Individual Factors on Raising the Nov 1978,P 47-49
Alternating Bending Fatigue Strength of 32. B. Vinokur, The Composition of the Solid So-
Case-Hardened Test Pieces, Hiirterei- lution Structure and Contact Fatigue of the
Technische Mitteilungen (HISI 6081), Vol 22, Case-Hardened Layer, Metall. Trans. A, Vol
Oct 1967,P 213-220 24, May 1993,p 1163-1168
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
96 / Carburizing: Microstructures and Properties

33. P. Barczy and J. Takacs, Endurance Improve- Cooled Carburized Microstructures, Conj:
ments of Planet Gear Pins by Heat Treatment, 1995 Carburizlng and Nitriding With Atmo-
paper 24, Heat Treatment '84, The Metals So- spheres, 6-8 Dec 1995, (Cleveland, Ohio),
ciety,1984 ASM International, 1995 p 71-76
34. G.T. Hahn, Y. Bhargava, C.A Rubin, and X. • A Inada, H. Yaguchi, and T. Inoue, Effects of
Leng, Analysis of the Effects of Hardened Retained Austenite on the Fatigue Properties
Layers on Rolling and Sliding Contact Perfor- of Carburized Steels, Kobelco Technol. Rev.,
mance, Conf. Proc. Processing and Perfor- Vo117, April 1994, p 49-51
mance (Lakewood, CO), ASM International,
• G. Krauss, Advanced Performance of Steel
Surfaces Modified by Carburizing, Conj:
35. C. Razim, Survey of Basic Facts of Designing
Heat & Surface '92, 17-20 Nov 1992 (Kyoto,
Case Hardened Components, unpublished pri-
Japan), Japan Technical Information Service,
vate print, 1980
36. B. Thoden and 1. Grosch, Crack Resistance of 1992, p 7-12
Carburised Steel under Bend Stress, Conj Proc. • G. Krauss, Microstructures and Properties of
Processing and Performance (Lakewood, CO), Carburized Steels, Heat Treating, Vol 4,
ASM International, 1983 ASM Handbook, ASM International, 1991,
37. V.K. Sharma, G.H. Walter, and D.H. Breen, p 363-375
The Effect of Alloying Elements on Case • F. Li and C. Li, The Influences of Heat Treat-
Toughness of Automotive Gear Steels, Heat ment After Carburizing and Retained Austen-
Treatment '87, The Metals Society, 1987 ite in the Carburized Layer on the Strength
38. G. Niemann and A Seitzinger, ''Calculation of and Toughness of Steel, Trans. Met. Heat
Scoring Resistance for Spur and Helical Treat. (China), Vol 6 (No.2), Dec 1985,
Gears," ISO 6336, International Standards Or- p 59-68
ganization, 1996 • J.D. Makinson, WN. Weins, and R.I. de
39. R.M. Matveevsky, Y.M. Sinaisky, and IA Angelis, The Substructure of Austenite and
Buyanovsky, Contributions to the Influence of Martensite Through a Carburized Surface,
Retained Austenite Content in Steels on the Conj: Advances in X-Ray Analysis, Vol 34,
Temperature Stability of Boundary Lubricant 30 July to 3 Aug 1990 (Steamboat Springs,
Layers in Friction, 1 Lubr. Technol. (Trans.
Colorado), Plenum Publishing Corporation,
ASME), July 1975, P 512-515
p 483-491
40. AG. Roberts, discussion appended to Ref
• D.L. Milam, Effect of Interrupted Cooling on
Retention of Austenite and Development of
4 W. Grew and A. Cameron, Role of Austen-
1. ite and Mineral Oil in Lubricant Failure, Case Hardness in a Carburizing Process,
Nature, Vol 217 (No. 5127), 3 Feb 1968, Conj: 1995 Carburizing and Nitriding With
p 481-482 Atmospheres, 6-8 Dec 1995 (Cleveland,
42. I.I. Kozlovskii, S.E. Manevskii, and I.I. Ohio), ASM International, p 111-116
Sokolov, Effect of Retained Austenite on • R.W Neu and H. Sehitoglu, Transformation
the Resistance to Scoring of Steel 20Kh2N4A, of Retained Austenite in Carburized 4320
Met. Sci. Heat Treat., (No.1), 1978, P 7{}-80 Steel, Metall. Trans. A, Vol 22 (No.7), July
43. S.E. Manevskii and I.I. Sokolov, Resistance to 1991, p 1491-1500
Seizing of Carburised and Carbon Nitrided • M. Przylecka, M. Kulka, and W Gestwa, Car-
Steels, Metalloved. Term. Obrab.Met., (No.4), burizing and Carbonitriding Bearing Steel
April 1977, P 66-68 (LHI5-521oo), Conj: Heat Treating: Equip-
44. 1. Terauchi and 1.1. Takehara, On the Effect of ment and Processes, 18-20 April 1994
Metal Structure on Scoring Limit, Bull. Jpn. (Schaumburg, Illinois), ASM International,
Soc. Meek Eng., Vol 21 (No. 152), Feb 1978, p 1994, p 233-238
324-332 • E. Shao, C. Wang, and Ci-S. Zheng, Im-
proving Rolling Contact Fatigue Strength for
SELECTED REFERENCES Carburizing Gears With Retained Austenite
and Carbide in Their Case and Its Applica-
• J. Grosch and O. Schwarz, Retained Austen- tion, Trans. Met. Heat Treat. (China), Vol 10
ite and Residual Stress Distribution in Deep (No.1), June 1989, p 31-43
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Retained Austenite I 97

• J. Siepak, Effect of Retained Austenite in (No.9), Sept 1997, P 69-71

Carburized Layers on Rolling Wear, Neue • D. Zhu, F.-X. Wang, Q.-G. Cai, M.-X. Zheng,
RuNe, Vol 29 (No. 12), Dec 1984, p 452-455 and Y.Q. Cheng, Effect of Retained Austenite
• B.B. Vinokur and A.L. Geller, The Effect of on Rolling Element Fatigue and Its Mecha-
Retained Austenite on Contact Fatigue in nism, Wear, Vol 105 (No.3), 1 Oct 1985,
Cr-Ni-W Carburized Steel, J. Met., Vol 49 p223-234
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Carburizing: Microstructures and Properties Copyright © 1999 ASM International®
Geoffrey Parrish, p 99-133 All rights reserved.
DOI: 10.1361/cmap1999p099

Chapter 5

Influential Microstructural Features

Microstructural features observed within and core of the carburized item were refined.
case-hardened steels include grain size, However, the advent of grain refinement by al-
microcracking, microsegregation, and nonmetal- loying obviated the need for carburized compo-
lic inclusions. Although these microstructural nents to be double quenched. Now, the majority
constituents are not considered in the same detail of heat treaters either employ direct quenching or
as those presented in the first four chapters, their single reheat quenching to harden their products.
significant influence on properties justifies their Figure 5.1, coupled with Table 5.1, illustrates
review. different thermal cycles used to effect hardening
and generally ensure a fine-grained product, pro-
vided the carburizing temperature for all pro-
Grain Size grams except D is not excessive. Much of the
flexibility for process cycles must be attributed
For optimal properties, it is essential that the to the availability of grain-refined steels.
grain size of a carburized and hardened compo- In addition to the process cycles shown in
nent is both uniform and fine. Generally, the Fig. 5.1, some cycles include a subcritical an-
starting point is a grain-refined steel having an nealing, or a high-temperature tempering opera-
ASTM grain size from No.5 to 8, though subse- tion, between the carburizing and hardening
quent mechanical and thermal processing can stages. By slow cooling from the carburizing tem-
change the final microstructure to be either more perature then annealing or tempering from 600 to
coarse or fine. 650°C, the part is rendered suitable for any inter-
At one time, grain size control was very much mediate machining needed. Another use of the
in the hands of the heat treater who, supplied high-temperature tempering involves a version of
with coarse-grained steels, employed double the double quenching program (D in Fig. 5.1).
quenching treatments to ensure that both case The first quenching operation is into a medium

_ ~rb~iz~g ~m~ra~re_ A B C D E F

r ~ ------ C'!""-@""-'l!£<'!'

l!! I C~~ra.!!.g~f~se

Carbon content,percent TIme- - - _

Fig. 5.1 Alternative heat treatment cycles for hardening carburized components. Seealso Table 5.1. Source: Ref 1
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
100 / Carburizing: Microstructures and Properties

Table5.1 Case and core characteristics resultingfrom the various heat treatments shown in Fig. 5.1
Treatment Case Core
A: Best adaptedto fine-grainedsteels Refined;excess carbide not dissolved; Unrefined; soft and machinable
distortion minimized
B: Bestadapted to fine-grainedsteels Slightly coarsened; some solution of Partially refined; stronger and tougher than A
excess carbide
C: Best adaptedto fine-grainedsteels Somewhatcoarsened; solution of excess Refined;maximumcore strengthand hardness;
carbide favored; austenite retention better combination of strength and ductility
promoted in highly alloyed steels thanB
D: Besttreatmentforcoarse-grained steels Refined; solution of excesscarbide favored; Refined; soft and machinable; high degree
austenite retention minimized of toughness and resistance to impact
E: Suitablefor fine-grainedsteelsonly Unrefined with excesscarbide dissolved; Unrefined but hardened
austenite retained; distortion minimized
F: Suitablefor fine-grainedsteelsonly Unrefined;excess carbide avoidedif combined Unrefined but hardened
with diffusion treatment; austenite retention
reduced; distortion minimized

Source: Ref I

held at a temperature not too far below the ACt • Boundary migration, which occurs at all tem-
from which the workpieces are then reheated for peratures but affects grain growth only at tem-
the second quench. The aim is to ensure grain re- peratures above 1100 DC
finement (Ref 2) and reduce distortion and the
risk of cracking. The first two mechanisms involve the decomposi-
tion of old boundaries, whereas the third involves
Evaluation of Grain Size the movement of boundaries.
Grain Size Control. Grain refinement by alloy-
Grain Growth. Vinograd et al. (Ref 3), who
ing is accomplished by adding certain elements
studied the behavior of twelve different steels
(e.g., aluminum or vanadium) to the molten steel
heated from 850 to 1250 °C, concluded that in the ladle after a thorough deoxidation treat-
grain growth takes place by three different mech- ment, usually with silicon. Silicon in adequate
anisms: quantities, although harmful in terms of internal
• The resorption of grains, which occurs at 50 oxidation during carburizing, is important as a
to 100 DC above the AC3 deoxidizer and, therefore, for grain size control
• The formation of new boundaries and grains (Fig. 5.2). Additions of aluminum and/or vana-
between 250 and 300 DC above the AC3 dium encourage the formation of compounds
(e.g., AIN or V4C 3) that, because of their ex-
treme fineness and relative stability, are able to
0.35 mechanically restrain grain boundary move-
21.4% coarsJ- 11.1 % coarse. Ie Rne-grained
~s ment during subsequent austenitizing treat-
! grainedcasts
x oarse-
grained_ ments. Of the grain refining agents, AIN is
I casts
more important for fine-grain stability. The ef-
I fect of grain refining treatments is limited, be-
-I • cause above a certain temperature within the
"if. I ~ austenite range, the precipitated compounds co-
•• •
-- x'wr- ~1
t •
alesce and then dissolve. Consequently, they
can no longer prevent grain boundary move-
•xx r
I •• ment. However, the grain coarsening tempera-

• x x
. I· x ••
ture is further increased by adding both alumi-
num and titanium. This addition is important
o 66.7% coarse. I
l when any high-temperature carburizing process
grained casts t I
• 18.2%coarse-
is being used, if indeed grain coarsening is a
I problem. The efficiency of aluminum to inhibit
0.05 austenite grain growth is illustrated in Fig. 5.3,
0.15 0.20 0.25 0.30 0.35
Carbon. %
which compares an untreated steel with an alu-
minum treated steel. Figure 5.3(a) shows that
Fi~. 5.2 Effect of carbon and silicon on grain size con-
troT of 665M17 steel made by basic electric-arc process the untreated steel coarsens progressively as the
with aluminum addition of 16 oz/ton. Source: Ref 4 temperature increases, whereas the treated steel
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 101

exhibits no serious grain growth until it reaches the coarsening temperature, and although con-
about 925 "C. Thereafter, the grains of the alumi- trolled hot working reduces the initial grain size,
num treated steel coarsen rapidly. Grain refining it nevertheless lowers the coarsening tempera-
agents other than aluminum and vanadium in- ture. Others (Ref 7) have suggested that an incor-
clude boron, titanium, niobium, cerium, and rect hot-working temperature contributes to the
sometimes chromium. occasional occurrence of coarse-grained alumi-
The duration of the austenitizing treatment has num treated case-hardening steels.
some significance with regard to grain coarsen- With respect to heat treatment, Kukareko (Ref
ing. For example, the steel referred to in Fig. 8), working with an 18KhNVA steel, related that
5.3(b) was fine grained when treated at 925 "C an increased heating rate to 930 "C results in the
for 1 h, whereas a 10 h treatment at that tempera- formation of a metastable fine-grained structure
ture causes the average grain size to rise. with a tendency to rapid grain growth via grain
Each steel has its own coarsening tendencies merging during subsequent isothermal heat
whether grain refined with aluminum or vana- treatment. On the other hand, a preliminary tem-
dium or left untreated, and steels containing al- pering operation and slow heating to the
loying elements, such as nickel and/or molybde- austenitizing temperature produces a stabilized
num, have a greater resistance to coarsening at grain structure.
conventional carburizing temperatures than In terms of modem carburizing practice, it is
plain-carbon steels. unlikely that conventional treatments at 925 "C
The mechanical and thermal histories of a part using satisfactorily grain-refined steels give rise
can influence the grain-coarsening temperature. to coarse austenite grain sizes. However, the
For example, Fig. 5.4 shows that prior normaliz- trend to higher carburizing temperatures, some-
ing lowers the grain coarsening temperature times in excess of 1000 "C, requires consider-
more than annealing does. As early as 1934, ation of all aspects of grain refinement (e.g., al-
Grossman (Ref 6) showed that repeated heating loying additions, hot working, and subsequent
to slightly above the AC3 temperature also lowers thermal treatments) to end up with a fine-grained

Temperature. °C
8 8


:::ii 2
-c 2
.~ 4
Cl 4

~ 6

10- 2 10- 1 10 102

Temperature. of Time. h
(8) (b)

Fig. 5.3 Relationship between austenitizing parameters and grain size for grain-refined and non-grain-refined AISI
1060 steel. (a) Effect of austenitizing temperature (2 h soak). (b) Effect of austenitizing time. Source: Ref 5
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
102 / Carburizing: Microstructures and Properties

product. Child (Ref 9) claimed that grain growth position products. If it is riecessary to determine
is not a problem with gas carburizing tempera- the austenite grain size, then the standard
tures up to 1000 °C. Nevertheless, Fig. 5.5 sug- McQuaid-Ehn test is employed. In the McQuaid-
gests that coarsening, at least in the leaner grades Ehn test, a steel sample is carburized in a me-
of carburizing steels, becomes likely at 1000 "C. dium of sufficient carbon potential to decorate
This observation is supported by Benito and the grain boundaries with continuous films of
Mahrus (Ref 11), who encountered an excessive cementite. The sample is then prepared for view-
increase in grain size when vacuum carburizing ing at l00x magnification, and the grain size
steels SAE 5115, 4320, and 8620 at 1000 to graded according to a table, such as Table 5.2, or
1100 °C. When four German steels were carbu- by comparison with examples, such as Fig. 5.7.
rized from 920 to 1100 °C (Ref 12), grain growth For case-hardening grades of steel, a grain size
was slight, although there was some local coars- of ASTM No.5 or finer (ASTM E 112) is speci-
ening at above 1020 °C in some samples. The fied.
chromium-manganese steels were found to be The austenite grain size influences the size of
more susceptible than the molybdenum-chro- the austenite decomposition products. The grain
mium steels. Thus, grain size is influenced by the size of the product is smaller than the austenite
chemical composition, the mechanical and ther- grain in which it grew. In the case of the
mal history, and the carburizing temperature. high-temperature transformation products (fer-
However, the reheat temperature is of particular rite, pearlite, or bainite), many new grains nu-
importance. Effective grain refinement of the cleate simultaneously along each austenite
case is achieved by reheat quenching from about grain boundary, and those reaching critical size
the Accrn temperature. first continue to grow at the expense of the oth-
Metallography. The austenite grain size of a ers. Thus, each austenite grain is replaced by
carburized and hardened steel is generally not several grains of the transformation product.
very obvious unless the carbon content is so high With respect to the diffusionless transformation
and the postcarburizing heat treatments such that that produces martensite, many faults develop
network carbides are visible during the metal- within each austenite grain during the quench.
lographic examination (Fig. 5.6). More often Each fault is a potential nucleation site from
than not, there are no network carbides, because which a martensite plate can grow; growth, when
they are regarded as detrimental and therefore it does occur, is rapid. The first formed plates are
avoided. Assessment of such a structure, at best, likely to be the largest and no longer than the
is qualitative (i.e., coarse, normal, or fine) and austenite grain diameter. Subsequently formed
made from the features of the austenite decom- plates then subdivide the remaining volumes of
the austenite.
Therefore, thinking in terms of carburized sur-
faces, the austenite grain size at the onset of the
900 quench influences martensite plate size (hence,
SAE3135 the microcrack content) and the size of the aus-
tenite volumes in the as-quenched product. Also,

'#. oor----------------.
~ Gas carnurized 8 h
·iii .g} 1 0 0 at 0.8%C potential
a. 50 t----+---~ <1) 2 .. 0 0 0 Coarse-grained CM70steels

j Cl
CD 0

~ ~ EN350series steels
• CM70F, CMBOF, CM90F steels

~ SAE8600series steels
25t---+-- :2 6
~ 7
CJ 8
~ 9
o --_....,,"" 10
11 '-:-1":"""":~~~.L--L--------1
1600 1700 1800 1900 2000 900 925 950 1000 1050
Case: Carburizingtemperature, 'F Temperature, 'C
Core: Healing temperature, 'F
Fig. 5.5 Grain-coarsening characteristics of eM series
Fig. 5.4 Reduction of grain-coarsening tempera- steels compared with conventional carburizing grades.
ture due to normalizing. Source: Ref 6 Source: Ref 10
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 103

Fig. 5.6 Prior austenite grain boundaries. (a) Coarse-grained SAE 1015 carbon
steel, carburized. 100x. (bl Fine-grained SAE 4615 nickel-molybdenum steel, carbu-

grain size affects the frequency and depth of in- grain size control and assessment by steelmakers
ternal oxidation and, therefore, the grain size of are sufficiently adequate to assure the customer
any high-temperature transformation products that fine-grained steels are being provided.
(HTTP) associated with it. The hardenability of a fine-grained steel is less
The austenitic grain size data for incoming than that of a coarse-grained steel having an
materials may be provided by the steel supplier identical chemical composition. The same ap-
or determined by the customer as part of routine plies to the case hardenability. However, whereas
acceptance testing. Unfortunately, grain size de- this distinction is especially meaningful for plain
terminations made on a small diameter forged carbon and very lean-alloy case-hardening
test bar may not truly reflect on those of the parts grades and may lead to problems during process-
that the test bar represents. For example, the ing, it is of less concern for the more alloyed
amount of hot working can be different, as too steels (Ref 14).
the heating and cooling rates during any simu- Francia (Ref 15), who provides a reminder that
lated carburizing treatments. Dietrich et al. (Ref hardenability is critical in gear steel selection,
13) warned how normalizing prior to grain size makes the point that hardenability should not be
testing could produce a false result and that slow sought by the use of coarse-grained materials.
heating (3 °Ctmin) to the testing temperature can Apart from having the potential to produce infe-
lower the content of coarse grains. Subcritical rior properties, coarse-grained steels distort more
annealing (710 "C) has a similar effect. Even so, than their fine-grained counterparts.

Table5.2 ASTM grain size

Gram dlamele!«), in.
I I 4.0 270 0.0113
2 2 5.6 340 0.0080
3 4 8.0 480 0.00567
4 8 11.3 679 0.00400
5 16 16.8 %1 0.00283
6 n 22.6 1360 0.00200
7 M 32.0 1920 0.00142
8 I~ 45.3 2720 0.00100

(a) Calculated for cub ical grains allOOx. (b) Average caleulated fora l4-sided solid of maximum ability for close packing. (e) Equ ivalent spherical
grain.not magnified
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
104 / Carburizing: Microstructures and Properties

Effed of Grain Size on Properties rect information is absent, other studies (of
non-case-hardened steels) are referenced to ob-
Most properties are modified by changes of
tain reasonable trends and guidance.
grain size, and some properties are adversely af-
Hardness. Austenitic grain size does not
fected by an increase of grain size. However, in-
greatly affect the surface hardness of carburized
formation dealing specifically with the effects of
and hardened surfaces and only marginally influ-
grain size on the properties of case-hardened
ences the core hardness. For high-carbon
parts is somewhat limited. Therefore, where di-
martensite, Kelly and Nutting (Ref 16) consid-
ered that grain size has a negligible influence on
the increase of hardness , and other factors, such
as carbon in solid solution within the twinned
martensite or carbide precipitation, are far more
significant. However, Zaccone et al. (Ref 17), us-
ing steels of varying chromium content and grain
size, showed with direct quenched samples that
for a given grain size the hardness decreases as
the retained austenite increases. With reheat
quenched samples in which both grain size and
retained austenite vary, there is a different trend
(Fig. 5.8). This trend difference might be attrib-
uted to grain size. The reheat quenched samples
were, as expected, finer grained than their direct
quenched counterparts.
For low-carbon ferritic steels (i.e., ferrite,
ferrite/pearlite, and the quenched and high-
temperature tempered microstructures), an ap-
proximately linear relationship between hardness
and grain size might be expected. The results for
three plain-carbon steels in the worked, quenched,
and high-temperature tempered condition are
shown in Fig. 5.9. The trend is for hardness to in-
crease as grain size decreases. However, over the
range of grain sizes typical for case-hardened
steels, the change in hardness is less than 20 HV.
Tensile Strength Properties. In heat-treated
steels, the tensile yield strength is influenced by
a number of factors including grain size (prior
austenite grain size or as-heat-treated grain size).
The yield strength is inversely proportional to the
square root of the austenite grain size (Ref 16).
This is illustrated in Fig. 5.10 for two martensitic
steels where the yield strength increases as the
grain size decreases by about 21 MPa per grain
size (d- I12, mm- I12) change . Other factors that
influence yield strength include carbon segrega-
tion, precipitates, and the substructure of the
martensite (dislocations , internal twinning) .
Schane (Ref 20), employing fine- and
coarse-grained SAE 1040 steels, showed that an
initially coarse-grained steel began to further
coarsen at a relatively low austenitizing tempera-
ture (810 to 840 DC), while an initially fine-
Fig. 5.7 Comparison of nominal ASTM 6 to 9 grain
size (with calculated grain size numbers of 6.08,7.13, grained material resisted coarsening until tem-
8.03 , and 8.97, respectively). Nital etch, 100x peratures above about 1000 DC were used. The
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features /105
/ 105

properties obtained
properties obtained after
after normalizing
normalizing at tempera-
tempera- again, macrostresses
macrostresses and microstresses
microstresses are in-
tures up to 1093
1093 °C °C are shown
shown in Fig.
Fig. 5.11. InIn volved.
terms of ductility,
ductility, the fine-grained
fine-grained steel
steel was
was the
the Fatigue Strength. The grain size of a steel af-
better of m thee two.
two. T h e inflections
The inflections in the curves
curves fects
fects its response to cyclic loading. Macherauch
coincide with the onset of grain
coincide with the onset grain coarsening. coarsening. (Ref 21) showed that, for a low-carbon
low-carbon steel,
Residual Acknowledging that
Residual Stresses. Acknowledging that grain coarse-grained
coarse-grained structures are inferior inferior to
differences might influence
size differences influence the the propor- fine-grained structureswhen
fine-grained structures whentested
tions of the transformation
transformation products
products in a case- ing fatigue conditions (Fig. 5.12). The same
hardened surface,
surface, particularly
particularly the plain-carbon
the plain-carbon
grades, then
then it can
can bbee reasoned
reasoned thatthat the
the residual
stress distribution
distribution within
within thethe carburized
carburized case will
case will
be modified
modified accordingly.
accordingly. Also,Also, coarse-grained
steels are
are more
more prone
prone toto distortion
distortion (which
(which is a re-
sidual stress effect)
effect) than fine-grained steels,
than fine-grained
which could
could bbee a reflection
reflection on on the influence
influence ofof 2260
grain size
size on
on hardenability.
hardenability. Coarse-grained
Coarse-grained steels
also tend to to b bee more prone to to cracking and and
microcracking during quenching or or grinding; >
Q) 220
2 20-


Rene at and
Reheat Di rect
Direct 180
que nch
quench que mch

59 ----------- -


58 ----------- -

— -...-;~::
57 f-- T7? J -------- --------- -
--, ,----
, , ----
* - - ,
- Fig. 5.9
6 10

5 . 9 Variation of
Source: Ref 1B
Grain size
18 22
(d" 1/2 ), mm-
size (d""1/2),
mnr 1l12

of hardness with ferrite grain size.


,, Grain size,
Grain size, ASTM number
ASTM number
V ,,
N S,
,, 3 5 7 99 11
35 13
13 15
, 2200 1
' ' — i —Ip ' II '
55 8650:0.5C-1Mn-0.5Ni-0.5Cr-0.25Mo

n 2000
865jV" e
o .......
275 ]!

o 20 40
40 60 80 g>
P 1800
250 g>
Austenite grain size, urn
• ~• /'
•g, 1600 0
225 ~

Retained austenhe,% CO

Cr,%.. Reheat IIDd
and queDl:h
quench DIrect
Direct quellCh
By = So + 3000d""1/2
o 1400 , , , 200
oo o0 25 23 ~
5- eft

1.3 32 32 o0 10
10 20 30

D 0.9 39 39 Grain size (d""1/2), mrrr1/2
(rf -1 ' 2 ), mm- 1/2

Fig. 5.8
Fig. 5 . 8 Hardness vs.
vs. grain
grain size.
size. Steel composition:
0.82C, 0.9Mn, 0.31 Si,
Si, 1.76Ni,
1.76Ni, 0.72Mo,
0.72Mo, and
and Cr
Cr (per
(per table).
table). Fig.
Fig. 5.10
5 . 1 0 Effect of
of prior-austenite grain size
size on
on the
Source: Ref 17 strength of
strength of martensite. Source:
Source: Ref
Ref 19
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
106 / Carburizing: Microstructures and Properties

Normalizingtemperature, °C NormaliZingtemperature, °C
760 871 982 1093 871 982 1093
110 60
........... ~
40 l!!
30 '0

~ 80
- - Coarse Qrain
•••• - Fine grain
... '
25 IS

20 L.-._--l._--==_ _--' 0:
~ 701----+--+----1 60 ...........
~ Yield
1--=--+--:-- pOint
.. -...
" ..
.. --
'. r---.... '.


"f4O,"=0--:-1600f=--18..l0-0--J20oo 1600 1800

Normalizing temperature. OF Normalizingtemperature. of
Fig. 5.11 Effect of austenitizing temperature (and grain size) on room tem-
perature properties of a 0.4% C steel. Solid line, coarse grain; dashed line, fine
grain. Source: Ref 20

trend has been observed with carburized steels retained austenite should be minimal and finely
(Ref 22). When reducing the grain diameter from dispersed. Such benefits are somewhat depend-
100 to 4 urn, more than doubled fatigue strength ent on the condition of the actual surface being
(Fig. 5.13). However, it is not clear what other cyclically stressed, as Fig. 5.15 shows (Ref 24).
factors are involved in achieving such an im- With coarser grained case-hardened surfaces,
provement. Pacheco and Krauss (Ref 23) illus- such as developed by direct quenching, the ten-
trated how grain size and retained austenite work dency is for crack initiation to be intergranular.
together to influence the fatigue strength (Fig. In fine-grained surfaces (e.g., developed by re-
5.14), although the individual contribution of heat quenching), crack initiation tends to be
each could not be isolated. For high bending fa- transgranular (Ref 25). Propagation is often
tigue strengths in carburized and hardened parts, transgranular, but the overload fracture can either
the martensite of the case should be fine and any be intergranular or transgranular depending on
the grain size or method of quenching (Ref 26).



E 200
106 }
Number of cycles
1: r::
}• ~ 200 ..c::
~ Fati9ue
=2' 20 -2'
ii ,......-' ~ l!!
100 ~I--"""'---+:=n~~--;

,......-' ~
f 15
o 2 4 6 8 10
150 f 50 0lI!l1tL---+---i---+---1

Decreasinggrain diameter (cr1l2). mm- l 12 0.1 0.2 0.3 0.4 0.5

Diameterof austenhegrains.IIm-l12
Fig. 5.12 Effect of grain size on the fatigue strength of a Fig. 5.13 Effect of grain size on the fatigue strength
0.11 % C mild steel. Source: Ref 21 of case-hardened test pieces. Source: Ref 22
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features/ 107

For instance, quenching from above the ACero size increases, the impact transition temperature
produces intergranular overload fractures. increases.
Bending and Impact Fracture Strength. Accord- The effect of grain size in relation to the im-
ing to Wang and Chen (Ref 27), in precracked pact strength of a plain-carbon steel in the nor-
fracture toughness specimens, crack propagation malized condition is shown in Fig. 5.11. The
occurs from a microcrack the size of a second- same trends apply to case-hardening steels in the
phase particle (e.g., a carbide precipitate). In blank carburized condition (see Table 5.3),
notched specimens, propagation occurs from a where the coarse-grained samples have higher
crack the size of a ferrite grain. Therefore, in im- tensile strengths, lower reductions of area, and
pact testing, grain size is more directly signifi- lower impact strengths than those of the
cant than it is for fracture toughness testing with fine-grained samples. Double quenching treat-
precracked specimens. ment makes for an improvement in each of these
In impact and fracture toughness testing, the properties. Table 5.4 considers a 0.45% C steel
ratio of intergranular to transgranular fracture where bending and impact strengths are highest
surfaces relates to the toughness of the test piece: when the grain size is at its smallest.
as the intergranular fracture ratio falls, the tough-
ness property increases (Ref 28). Further, a small
grain size appears to equate to a reduced inter-
granular fracture ratio. Therefore, the toughness Microcracking
of a case-hardened surface is dictated by the
intergranular strength. The finer the grain size is, A structural feature sometimes observed in
the lower the intergranular fracture ratio is, quenched bearing steels that can develop in car-
hence the higher the toughness is. When the burized and hardened surfaces is microcracking
transgranular mode of fracture takes over, the associated with the martensite plates. Such
presence of retained austenite in the structure be- cracks run either across the martensite plate or
comes more significant with respect to tough- along the side of the plate (Fig. 5.16), that is,
ness. A point to bear in mind is that as the grain along the interface separating the martensite and
the adjacent austenite. Microcracks have also
been observed at the prior austenite grain bound-
aries (Ref 30). Detection requires careful prepa-
Grain size, ASTM scale
ration of the metallographic sample and only a
8 9 10 1112
1800..----.... --.,...----r----, light 2% nital etch immediately prior to viewing
at 500 to l000x magnification.
• Plasmacarburizing
• Gas carburizing Microcracks are formed when the strains gen-
erated at the tip of a growing martensite plate are
1600 sufficient to induce cracking in any plate or
boundary with which the growing plate im-

~ 1400
~ ~
~ 203~ Ql
:::l . (0.08 mm removed)
.2> 2000
,g'1200 ~,.,..._ 1
If. ;f ~ 1500 ~ :--...{ Coarse-grained
~ 1200

~ 800
. .t . . . '.. . ;h_._._._.
- .... ' ..... ,.....
(0.08 mm removed)

Coarse-grained - .... ' . :-.....
1000 ~600 ............
400 Fine-grained - ~

300 Steel 16MnCrS

800L-_ _..I.... _ _- I -_ _--I-_ _.....J145 2ooL.._L--..L.::--l..:--.L;---7---:~~
1 10 102 1()3 1()4 105 106 107
0.20 0.25 0.30 0.35 0.40
Numberof cycles
Grain size, (u-1I2), mm-1I2
Fig. 5.15 Effect of grain size on fatigue strength of
Fig. 5.14 Fatigue limits of plasma and gas-carburized case-hardened gears with and without removal of a 0.08
specimens as a function of austenitic grain size. A, re- mm surface layer. Coarse-grained steel, ASTM 1-4;
tained austenite. Source: Ref 23 fine-grained steel, ASTM 6-8. Source: Ref 24
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
108 / Carburizing: Microstructures and Properties

TableS.3 Effect of grain size on the core properties of case-hardened steels

Mal lmum lensOe
MeQuaid-Ehn lest. Izodimpaet Reductioo 01 stress
Steel Heallrealmenl Grain ASTM grainsize 1I·1b kr/cJD1 area, % 10os/iD.l kI/IIIli?
Carbon Pot quench Fine 8 42 7.2 54 46 72
Coarse 2-3 16 2.8 39 49 77
Single quench Fine 8 58 10.0 62 37 58
Coarse 2- 3 13 2.2 45 41 65
Double quench Fine 8 99 17.1 69 37 58
Coarse 2-3 52 9.0 59 38 60
Nicb:l-molybdenum Pot quench Fine 7~ 96 16.6 64 51 80
Coarse 3 31 5.4 54 60 94
Single quench Fine 7~ 87 15.0 62 51 80
Coarse 3 40 6.9 61 56 88
Double quench Fine 7~ 93 16.1 64 48 76
Coarse 3 72 12.4 56 49 77

Source:Ref I

TableS.4 Deterioration of properties with increasinggrain size

Sohation Bend Impact Bend angle LencthoC M u imum Austenitecraio
lemperature. strenctb, Deflection, Hardn.... (notcbed). orrtactured martens;.eneedles, Ienctb oC size(Gust scale).
"C krImmZ mm IIKC Smooth Notched kr/cmZ sampte (Gust scale).vade martensite, lim vade
85O(a) 361 7.12 53 4.2 0.6 2 20"30' 7~

950 414 7.97 57 12 0.7 2.6 50"30' 8-9 12-16 9--10

1050 397 5.57 57 II 0.6 2 48°30' 9--10 16-20 8-9
1150 385 5.46 56 6 0.6 1.9 26°00' >20 8-6

Steel 45. induction hardenedand tempered. 160-170 °C. <a) Someferritewith martensite. Source: Ref 29

pinges. The growing plate itself may be cracked Factors Influencing Microcracking
by the impingement (Ref 31). A crack must al-
ways involve an impingement , although not all Carbon Content of the Steel. Microcracks form
impingements cause cracks.
only in plate, not in lath, martensite. For
iron-carbon alloys, plate martensite starts to ap-
pear when the carbon content is above about
0.6%, as shown in Fig. 5.17 (Ref 26). Although
microcracking is not observed in steels with car-
bon levels of less than the eutectoid carbon
(0.78% C) (Ref 32), above th at value the
microcracking incidence increases with carbon
content. Con sequently, in case-hardened sur-
faces, microcracking may be unavoidable, be-
cause surface carbon contents often exceed the
eutectoid value. Further, as the alloy content in-
creases, the eutectoid carbon content decreases
and could fall below 0.6% for a 3%Ni-Cr steel.
Provided that quenching of the austenitized
test pieces is adequate, the final microstructure
consists of martensite and austenite. The quan-
tity of austenite largely depends on the M, and
M f temperatures, which are strongly influenced
Fig. 5.16 Microcracking in a Ni-Cr steel that also ex- by the carbon content. The retained austenite
hibits microsegregat ion. 1000x content is unlikely to have much bearing on the
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 109

Austenite (y) 1000

!f- ~
Ferrite (a) + carbide (C) 600 ::;]l!F
i§ i§
c. 800 c.
E 400 E
{!? {!?

400 200
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Carbon, wt"10

Fig. 5.17 Morphological classification of martensite in Fe-C alloys. Source: Ref 26

microcrack count, which has more to do with cracks. In the long, slender {225} martensite
how much martensite is present. plates, the impingements are generally colli-
Most of the studies on the subject of sions. However, the habit plane is of secondary
microcracking have used steel with a carbon importance in microcrack formation (Ref 34).
content in excess of 1.0% and an inherently The primary factors are the length of the
coarse grain size. In contrast, carburized steels martensite or, perhaps, a high length to width ra-
generally have surface carbon contents of less tio at the instant of impingement (Ref 34). As the
than 1.0% and are grain refined. martensite plate thickens, there is a tendency for
Carbon in the Martensite. The temperature and the cracks to be along the interface.
duration of austenitizing determines how much Effect of Alloying Elements. It has been suggested
of the carbon is dissolved and how much of it re- that microcracking is more likely to occur in
mains combined as carbides. With respect to
microcracking, it is the carbon in solution in the
martensite that is important (Ref 33). Figure 5.17 25
o Ref33
shows the Accm for a Fe-1.39%Calloy is -920°C. • Ref37
On heating from the ACl (-710 "C) up to 920°C,
the amount of carbon going into solution in- .. 20
creases progressively. At and above 920 °C, fol- E
lowing a reasonable soak, all the carbon is in so-
lution. Quenching the steel from any temperature
~ 15
within this range determines the degree of solu-
tion and, hence, the extent of the microcracking ~

(Fig. 5.18). Once full solution is attained, there is CD
little change to the microcracking sensitivity. ::>
Increasing the carbon content of the martensite
increases the tetragonal distortion and, therefore,
the brittleness of the martensite (Ref 33). It also
changes the morphology of the martensite from
one with a {225} habit plane to a {259} habit
plane. Nickel favors the {259} type, whereas
U 5 )IJ Fe-1.39%C

chromium and manganese stabilize the {225}

habit planes. The {259} martensite forms in o 1400 t600 1800 2000 2200 OF
bursts by autocatalytic action; therefore, many of 760 871 982 1093 1204°C
Austenitizing temperature
the impingements are not collisions. Any
microcracks associated with this type of Fig. 5.18 Effect of austenitizing temperature on micro-
martensite tend to be interface or boundary crack sensitivity. Source: Ref 33,37
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
110 / Carburizing:
110 Carburizing: Microstructures and
and Properties

steels for which the major alloying elements are

steels are it might be be expected that microcrack
microcrack incidence
the carbide formers (Ref 35). On the other hand, hand, increases with austenite
increases austenite grain size,
size, up to aa point
Davies and Magee
Magee (Ref 34) do do not regard alloy- (to 200
(to 200 urn
u,m grain
grain diameter
diameter according
according to to Ref
Ref 34).
ing elements (Ni,(Ni, Cr, Mn)
Mn) asas having aa direct in- Rauch and
Rauch and Thurtle
Thurtle (Ref
(Ref 32)
32) were
were inin agreement,
fluence on
fluence on microcracking.
microcracking. Indirectly,
Indirectly, though,
though, al-
al- stating that
stating that both
both thethe frequency
frequency and and thethe size
size of
loying elements
loying elements do do alter
alter the
the martensite
martensite habit
habit microcracks increased with grain
microcracks increased with grain size (Fig. size (Fig.
plane of
plane of high-carbon
high-carbon materials
materials andand can
can thereby
thereby 5.19). Brobst
Brobst and and Krauss
Krauss (Ref
(Ref 30),
30), studying
studying aa
affect the microcrack event. In
affect the microcrack event. In commerciallycommercially Fe-1.22%C alloy, also confirmed that there were
Fe-l.22%C alloy, also confirmed that there
heat treated
heat treated parts,
parts, bearing
bearing steel
steel 52100
52100 andand case-
case- fewer cracks in the smaller grains, but for those
hardened, coarse-grained,
coarse-grained, SAE SAE 8620
8620 steels
steels have
have fewer cracks in the smaller grains, but for those
hardened, cracks which
cracks which did did form,
form, the
the proportion
proportion of of grain
received the
received the most
most attention.
attention. Neither
Neither of of these
boundary cracks
boundary cracks hadhad increased.
steels is
steels is particularly
particularly well
well alloyed;
alloyed; there
there are
are many
case-hardening grades more heavily alloyed, yet Quench Severity. Kern (Ref 35) suggested that
case-hardening grades more heavily alloyed, yet microcrack
the microcracking
microcracking incidence
incidence seems
seems to to be
be nono vigorous quenching contributes to microcrack
worse. What
What these
these two
two steels
steels can
can have
have inin com-
com- formation. However, this conclusion is not sup-
mon is is the
the carbon
carbon content,
content, and
and ofof the
the elements
elements ported by laboratory tests in which water water
added to steels, carbon is the most potent with quenching, oil quenching, or simulated simulated
added to steels, carbon is the most potent with
respect to
respect to microcracking.
microcracking. martempering are employed (Ref 37) or where
Plate Size andand Grain
Grain Size. Although micro- quenching into 3% 3% caustic solution, oil at 60 °C,
cracks are known to occur in small martensite or a salt bath at
or a salt bath at 180°C 180 °C areare utilized
utilized (Ref(Ref 32).
frequently ob-
plates (Ref 32, 36), they are most frequently Microcracks formed
Microcracks formed regardless
regardless of of thethe type
type ofof
served in the larger plates (Ref 32, 34). The high- quenching used,
quenching used, thereby
thereby implying
implying thatthat quench
est density is associated with the longest and severity is
severity is not
not significant.
thinnest plates, presumably because the larger Tempering. Microcracking in case-hardened case-hardened
plates are likely to be subjected to more im- surfaces may be aggravated by the presence of
pingements by other large plates striking them hydrogen, which is absorbed during carburizing
at the necessary velocity and tip strain energy and reheating in an endothermic atmosphere.
(Ref 34). The maximum plate size, or volume, is Welding research indicates that the micro-
directly related
directly related toto the
the grain
grain size
size and,
and, therefore,
therefore, cracking of martensite in the heat affected affected zone
influenced by the presence of hydro-
of a weld is influenced
gen (Ref 38). For example, the martensite of a
9 p weld cooled to and soaked at temperatures above
16 150 °C is is free
free from
from microcracking,
microcracking, whereas whereas aa
,I / comparable weld
comparable weld cooled
cooled toto below
below 130°C
130 °C con- con-
14 tains microcracks
tains microcracks even even when
when soaked
soaked at at that
that tem-

ooxx 12
1018 rimmed
steel ~
II perature. When
area, major
When microcracks
microcracks exist
exist in
area, major cracks propagate from them due to
in the
the weld
due to

,I Z
§o the action of low applied loads and
the action of low applied loads and delayed hy- delayed hy-

drogen cracking
cracking (Ref(Ref 39).
~22 Si-killed
11 coarse grain
Conventionally carburized
carburized and freshly
ched parts contain hydrogen, may contain
freshly quen-

microcracks, and are somewhat brittle. It would

/ O ~arbUriJed
0 Carburized or or
carbonitrided at
carbonitrided at _
seem prudent, therefore,
therefore, to temper components
£ 6 900°C soon after
after quenching to drive off the hydrogen

\ /; +
900 °C
Carburized or
•+ Carburized
843 °C
at -
and reduce the risk of delayed cracking. Immedi-
ate tempering would also induce some measure

, ,
(III V 843
V 788
refined at
Grain°Crefined at
788 °C
of microstress relaxation to toughen
toughen the martensite/
austenite structure. Tempering has an added ben-
efit of causing the smaller microcracks to heal

*,~<JV 7 6
1018 Al-treated
fine wain
-fine grain
5 4 3 2
(Ref 40-42),
4Q--42), and tempering at temperatures as
low as 180°C
180 °C for as little as 20 min has been
Grain size, ASTMscale
size. ASTM scale shown to have a marked effect effect (Fig. 5.20).
Fig. 55.19
. 1 9 Influence
Influence of
of grain
grain size
size on
on microcrack
microcrack fre-
fre- Healing is attributed
attributed to volume changes and as-
quency inin some
some carburized
carburized steels.
steels. Source:
Source: Ref
Ref 32
32 sociated
sociated plastic flowflow (Ref 42).
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 111

Effect of Microcracking on Properties. Micro- consider, it is difficult to assess the contribution

cracking has attracted attention in terms of of anyone of them. Even so, Apple and Krauss
causes and contributing factors, but few studies, (Ref 43), through examining the data including
as yet, have been carried out to determine the ef- fracture characteristics, concluded that the dif-
fect of microcracking on the more important ma- ference between the fatigue strength of the di-
terial properties of heat-treated bearing steels or rect quenched and reheat quenched samples
case-hardened steels. Occasionally, premature may be attributed to the differences in micro-
failures have been attributed to the presence of crack distribution and size. Also, the absence of
microcracks, but in failure analyses, there are microcracks at the immediate surface of the dou-
generally so many variables involved that it is ble reheated samples may, in part, account for
difficult to isolate the primary contributor to fail- the relatively superior test results. In viewing
ure. these results, one might speculate that grain size
The presence of microcracks implies that and hardness accounted for about 80% of the dif-
structural microstresses have been relieved by ference in fatigue results, and microcracking ac-
microcrack formation. However, now the cracks counts for the rest.
must be regarded as stress concentrators, and the Of the three heat-treated conditions (Ref 43),
following questions apply. Are the cracks large the largest microcracks were observed in the
enough to be significant as initiators of service double quenched samples, although the crack se-
cracks? Do they generally reside in locations of verity was the least. However, the overall maxi-
maximum potency? The cracks are mostly lo- mum crack severity was not at the surface but
cated within the confines of the prior austenite about 0.1 to 0.2 mm beneath the surface.
grain. If a service crack initiates at a grain Panhans and Fournelle (Ref 44) reported the
boundary and starts to propagate along the grain
boundaries, most of the microcracks are out of
harm's way. When a service crack is, or be- Tempering temperature. °C

comes, transgranular, microcracks would likely AQ 100 200 300 400 500 600 700

hasten its propagation. As yet there is no clear 2000

evidence of just how detrimental microcracks are "/0

to the service life of a component. Most ~ 1000

microcracks are probably not close enough to the TI


surface and not of critical defect proportions to '0 500

initiate a trans granular service crack. Neverthe-

400 •
E 300 ,,
less, a crack is a crack, and it should be regarded Z ,
with some mistrust.
If it is assumed that cracks will be closed by 100 l..-.L---'---'--'_L.--->----'-----'----'_-'--'-------'---.J
compressive macrostresses and opened by ten- AQ 200 300400500 600 700 BOO 9001000110012001300
sile macrostresses, then the influence of micro- <a>
cracks could vary according to situation. Carbu-
rized and hardened cases contain residual com-
14 10min 1h 10 h
pressive macrostresses, which reduce any known
or unknown adverse traits that microcracks Fe-l.2C

might have. Through-hardened high-carbon o lBO·C

• 200·C
steels, on the other hand, might have surfaces in • 225·C
which the residual macrostresses are decidedly • As quenched

tensile after quenching. The macrostresses in this

instance add to any adverse influence of the
Influence on Fatigue. Apple and Krauss (Ref
43) investigated the influence of microcracking o Ll0"'2~~~.....L.,..-~~~.u.J..,-~~~",":-<--'--'
on the fatigue resistance of case-hardened sam- (b) Time, S
ples, using different quenching treatments to
vary the microcrack population. The quenching Fig. 5.20 Influence of tempering on microcracking.
(a) Effect of tempering temperature on the number of
treatments led to differences in grain size, hard- cracks per unit volume. (b) Effect of tempering tempera-
ness, retained austenite content, and microcrack ture and time on SV' microcrack area per unit volume of
density (Fig. 5.21). With so many variables to specimen. Source: Ref 41,42
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
112 / Carburizing: Microstructures and Properties

same findings. Removal of a surface layer, by creased surface carbon content. When a steel is
whatever means, could actually bring the zone of carburized from 925 to 935°C and below the
maximum crack incidence to the surface. conditions necessary to promote grain coarsen-
Control of Microcracking. The use of grain- ing, the resultant grain size is likely to be normal
controlled steels and the control of carburizing for that quench. The extent of the microcracking
condition to limit surface carbon content help re- should be less than it is for the high-temperature
duce the likelihood of microcrack formation. treatment. Again, an increase of the carbon con-
The quenching conditions then become important tent could lead to more microcracks.
to ensure that the grain size remains fine. Quenching from below the Ac cm produces a
High-temperature carburizing at greater than finer grain size and forms carbides in high-carbon
1000 °C can produce coarse-grained or mixed surfaces (or the lack of carbide solution during re-
coarse-and-fine-grained microstructures, be- heating for quenching). If some of the carbon is
cause the grain-coarsening temperature has been tied up as carbides, then the matrix material has a
exceeded. If grain size and microcracking are re- lower carbon content and, hence, a reduced risk of
lated, a high incidence of microcracking might microcracking during quenching. In reheat
be expected when quenching a coarse-grained quenched materials, mierocracking is rare (Ref
material from above the ACcm or with an in- 26). However, when considering a lower quench-

Distance below surface. mm

0.5 1.0 1.5 2.0 2.5
900 I I 10
3 65 'I
800 E
60 .~
IIIG> 700
"E 55 ~
600 :J:

8c 50 ~ '2
~ 500 ~ ::>
45 8 .,

400 40 "'~"



35 Doublereheat


Single, A

reheat \


... ~o
o 0
Directquench ---Q.g-.....-".....:1.-_
5 0

Fig. 5.21 Hardness, retained austenite, microcrack density, and bending fatigue curves for carburized and hardened
SAE 8620 steel quenched by three methods: direct quench, ASTM 1-3 grain size; single reheat, ASTM 4-5 grain size;
double reheat. Source: Ref43
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features /113

ing temperature to control microcracking, it is specification range. What is of more concern is

necessary to be mindful of the effect that the variability that occurs on a microscale over
low-temperature quenching will have on the core very small distances. This variability is referred
microstructure. If ferrite is unacceptable in the to as microsegregation.
core, then there is a limit to how low the quench-
ing temperature can be reduced. Formation of Microsegregation
Krauss (Ref 26) reported that quenching an
initially coarse-grained steel from below the Ingot Solidification. Microsegregation of steels
Ac crn causes fine microstructural features, few arises during the solidification process. In this
microcracks, and that on testing the overload section, only microsegregation caused by solidi-
fractures were trans granular. Quenching from fication following casting is discussed. Subse-
above the ACcm produces a coarse-grained struc- quent segregation of elements during heat treat-
ture, microcracks, and predominantly intergranu- ments, such as the segregation of phosphorus to
lar overload fractures. austenite grain boundaries during austenitizing
The rate of quenching has already been dis- prior to quenching, is not addressed here.
cussed; it has little significance on microcrack When a steel is cast into the ingot mold, the
severity. Interrupted quenching, on the other first material to solidify is that adjacent to the
hand, is very significant. Lyman (Ref 46), inves- cooler mold walls. If there is no special prepara-
tigating microcracking in AISI 52100 steel, tion applied to the mold walls to slow the cooling
quenched the austenitized samples into oil held rate, then the first metal to solidify, called a
at a temperature just below the M, (138°C) of "chilled" surface layer, is a thin layer of small
the steel. After an appropriate dwell, the samples equiaxed crystals with the same composition as
the liquid metal. At the inner surface of the
were transferred into a salt bath held at 260 °C
and then quenched off. The aim was to temper, chilled layer, the situation is energetically favor-
stress relieve, and toughen the first formed able for crystals above a critical size to continue
growing inward. The growth is cellular initially
martensite. The results are as follows:
but soon gives way to dendritic growth parallel
• Without the 260 °C temper, simply an arrest to the thermal gradient. The dendrites grow long
at just below the M, before continuing the (see the columnar zone, Fig. 5.22 and 5.23), but
quench, microcracking was severe. such growth slows as the thermal gradient be-
• When the sub-M, arrest prior to 260°C tem- comes flatter. Meanwhile, in the liquid ahead of
per produced -10 to 20% martensite, only a the thermal gradient when the temperature has
few microcracks formed. fallen sufficiently, nucleation takes place at many
• When the martensite formed prior to temper- sites ( i.e., on suitable substrates, such as nonme-
ing was -30 to 40%, no microcracks formed. tallic inclusions), followed by uniform dendritic
growth at each site until solidification terminates
Clearly, there are a number of ways of controlling
microcracking, including tempering, which can
heal the smaller microcracks. Sinkhead

Maximum r~:=;l::J.--Shrinkage
positive cavity
Ingot shell
equiaxed segregation
For design purposes, it is generally assumed structure Columnar
that steels are homogenous. Unfortunately, they structure
are not. Macrochemical analysis surveys of in- Dendritic--\:;~ ~:-':;';~-High
structure segregation
gots cast from a single ladle of molten steel re- and center
veal differences of composition. These surveys porosity
Inverted "V"
also show that composition varies from the bot- "A" segregation Negative
tom to the top and from the center to the outside segregation
Sedimentary equiaxed
of each ingot. A batch of forgings made from a structure
single ingot will therefore exhibit part-to-part contamination
composition variability, as well as variations inclusions
within each part. However, such variations do Fig. 5.22 Schematic illustrating macrosegregation
not generally depart from the intended steel in a large steel ingot
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
114 / Carburizing: Microstructures and Properties

when neighboring grains impinge. These grains equiaxed growth varies depending on the length
constitute the equiaxed grained zone at the center and steepness of the thermal gradient. Thus, the
of the ingot. mode of solidification is essentially columnar for
The Development of Microsegregation. An al- continuously cast steels and equiaxed for very
loy of composition Xo is presented in the hypo- large conventionally cast ingots.
thetical constitution diagram of Fig. 5.24. When Microsegregation Tendencies of Alloying Ele-
the temperature of the molten metal falls to the ments. Steels are complex alloys containing
liquidus at Xl, solid material of composition many elements. Some elements are intentional
A 1 nucleates within the melt. With further cool- and wanted, and others are residual and possibly
ing, the composition of any growing crystal unwanted. Each element has its own segregation
changes from A 1 at its center through to A4 on tendency, although in some instances the segre-
its surface. Meanwhile, the composition of the gation tendency of one element can be modified
surrounding liquid changes from X 1 to X 4 , by the presence of others. For example, silicon
which is the last of the liquid to solidify. There- and manganese influence the segregation of mo-
fore, there is a distinct difference in impurity or lybdenum, and manganese influences that of sul-
solute content between the first and the last ma- fur. The generally accepted order of the suscepti-
terials to solidify. This difference is what is bility of elements to segregation is, from most
called ''microsegregation.'' Figure 5.25 shows an prone to least: sulfur, niobium, phosphorus, tin,
example of microsegregation in which the com- arsenic, molybdenum, chromium, silicon, man-
position of a cross section of a dendrite is re- ganese, and nickel. Table 5.5 provides a basis of
corded (Ref 47). It shows a comparatively low comparison for some of the common elements.
solute content, in this case nickel and chromium, Table 5.5 clearly indicates that, when solidifica-
at the center of the dendrite. tion has been directional (columnar growth), the
As the columnar grains grow forward to- intensity of segregation is less than in the central,
gether from the surface (Fig. 5.23), they push equiaxed grain zone.
the more impure liquid ahead of them toward The Effect of Mechanical and Thermal Treat-
the central zone. Therefore, the central zone ments. Case-hardening steels are generally sup-
contains more of the impure material than the plied in the wrought condition, that is hot rolled
columnar zone. When the central zone has al- or forged to some convenient shape and size.
most solidified, the least pure liquid, including During these hot-working processes, the micro-
the lower melting point nonmetallic inclu- segregated areas are given a directionality related
sions, has nowhere to go except to the inter-
faces between the impinging dendrites. Conse-
quently, on final solidification, the central zone ·Xo
is the more heavily macro- and microsegregated
(Fig. 5.26). Laren and Fredriksson (Ref 47) ob-
served that the maximum microsegregation in 1
and 9 ton ingots occurs at three-fourths of the
distance from the surface to the center of the ingot.
Steels solidify in the manner described, al-
though the proportion of columnar growth to


grains -

Equiaxed % impurity --+

grains I
Composition of first Composition of last
metal to solidify metalto solidify
Fig. 5.24 Schematic constitutional diagram of how
Fig. 5.23 Section through a conventionally cast ingot compositional variations develop during solidification
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features /115

to the amount and direction of working. Not only Additional mechanical and/or thermal treat-
that, the intensity of microsegregation is reduced ments to remove or reduce microsegregation add
somewhat, as indicated in Table 5.6 for chro- to the cost of manufacture. Unless there is a
mium and nickel segregation. clear-cut reason to have a homogenous material,
A high degree of homogenization can be ef- then perhaps things are best left as they are.
fected thermally by soaking the segregated ma- Macrostructures and Microstructures. The
terial at an elevated temperature. However, the distribution of microsegregation in wrought
soak times to ensure virtual complete homoge- steels depends on how much working has been
neity can be very long, particularly below 1200 done to shape the part. In a large forging with
°C (Fig. 5.27). Nevertheless, a significant reduc- little deformation, the cast structure is not
tion of microsegregation can be achieved by a eliminated, and much of the dendritic form of
balanced combination of mechanical and segregation persists (Fig. 5.28). In small
high-temperature heat treatments. forgings, on the other hand, the metal is gener-

Path A-A


<ft. ., " Ni

I "

1.5 I

E 1.4 I

V\\ . , ,




1.3 '"
I J i
\ I 1\

B B 1.2

o 100 200 300 400 500 600
Path transverse distance, 11m

Path B-B Path C-C
1.7 1.7

1.6 \ 'I\cr
\ I

I · .....

, , \ <ft.
" ,,
.: 1.5
o ., ··,. , ~

1\ ·... ,
" Ni
E 1.4
' ,
....,: f\ " ... "'., Ni

o I

• I
o I , I


\ ... .11
• I

1.3 1.3

1.2 \/, 1.2

\ 1/ 1\( cr\


1. 1 1. 1
o 100 200 300 400 500 600 0 100 200 300 400 500 600
Path transversedistance, IlITI Path transverse distance,11m
Fig. 5.25 Chromium and nickel segregation around a dendrite cross. Melt composition: O.36C, 0.35Si,
O.68Mn, 1.48Cr, 1.44Ni, O.20Mo. Source: Ref 47
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
116 / Carburizing: Microstructures and Properties

ally more severely worked. While segregation, Reaustenitizing and fast cooling in the pearlit-
albeit less intense, survives the mechanical treat- ic range causes the carbon to redistribute fairly
ment, it is redistributed as alloy-rich and al- evenly to give the impression that the segregation
loy-lean bands flowing in directions dictated by has been removed. Unfortunately, the alloy seg-
the working process (Fig. 5.29, 5.30). regation will still be there, essentially unaffected.
In microsegregated materials, the heat treat- It is merely masked by the redistributed carbon.
ment response of the alloy-rich areas is different In the surface of a carburized case where the car-
from that of the alloy-lean areas. Consequently, bon may be evenly distributed, the presence of
it is possible for each to transform at different microsegregation might be detected by observ-
times and to different products on cooling from ing alternate areas of martensite and bainite or of
the austenitizing temperature. In low- and medium-
carbon steels, austenitized and slow cooled, the
microsegregation is identified as alternating ar- Table 5.5 Segregation indices for some elements
eas of ferrite and pearlite (banding). in steel
Segregation index (lS)(a)
Alloyingelement Columnar crystals Equiaxed c!JStais
Chromium 1.31-1.38 1.47-2.10
Nickel 1.06-1.24 1.10-1.50
Manganese 1.07-1.27 0.40-1.95
Molybdenum 1.70-2.00 1.9-7.3
Niobium >2000
Phosphorus 2.05 >170
Arsenic 1.23 -75

(a) (Concentration at interdendritic site)/(concentration at dendritic

core). Source:Ref 48,49

Table 5.6 Changes in microsegregation intensity

due to hot working between 1200 and 1000 °C
Segregation index (IS)
o 100 200 300
__ 400
Columnar crystals
Equiaxed crystals
Distance from ingotsurface. mm Chromium(as cast) 1.15 1.16 1.72 1.54 1.35 1.69
Chromium(98%reduction) 1.08 1.12 1.45 1.20 1.14 1.46
Nickel (as cast) 1.12 1.08 1.40 1.30 1.27 1.42
Nickel (98% reduction) 1.02 1.08 1.39 1.29 1.05 1.39
1 ton
Source: Ref 50



'#. 3
E ~
0> 0.8
.~ .,

o 2 'E
I.Y~ '0
Crmin ~
T T Ci
1 TJ.t-:
J. J. E

U 0.2
o 100 200 300 400
100 200 300 400 500
Distance from ingot surface. mm Soaktime, h
Fig. 5.26 Maximum and minimum chromium con- Fig. 5.27 Effect of homogenization temperature and
centrations asa function of distance from surface. Source: time on the intensity of manganese segregation in an
Ref 47 EN39 steel. Source: Ref 51
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
austenite and martensite, depending on the actual amounts of pearlite and ferrite with the slower
compositions and the cooling rates involved. cooling rates. Figure 5.3I(a) relates these micro-
Where the tempered martensite normally etches structures to the hardness.
dark, a light 2% nital etch should determine if Microsegrcgation influences the hardness of
significant microsegregation is present. the case, particularly if it leads to local concen-
trations of retained austenite or bainite in the
Effects of Microsegregation on Properties mainly martensitic structure. Both austenite and
bainite are softer than the martensite at a given
Influence on Hardness. Figure 5.31 (b-d) carbon level.
shows an example of the variations in micro- Influence on Tensile and Toughness Properties.
structure that can result from the presence of The tensile strength and yield strength are un-
microsegregation; the cooling rates are those of likely to be affected by the presence or absence
an end-quenched hardenability bar. Within this of microsegregation. The ductility indicators, on
range of cooling rates, the alloy-rich areas have the other hand, will be affected. Kroneis et al.
transformed to either martensite or bainite. (Ref 53) showed that the reduction of area of a
Meanwhile, the alloy-lean areas have trans- conventionally melted and forged medium-carbon
formed to martensite and bainite with increasing alloy steel (5: I to 10:I reductions) was about

Fig. 5.29 Sect io n throu gh a for ged slid ing clut ch hub.

0 Zone of max segregation

12Cll 200 Zone
Fig. 5.28 Dendritic microsegregation in a fractured I segregation Homogeneous
gear tooth.2x zone
Zone of max segregation

Zone of min segregation

I a. 100
Homogeneous zone

5 !artensite V-
Ferrite Pearlite
U 50
Cii a 1/
a 20 40
60 80 100
Distancefrom end face, mm
Fig. 5.30 Microstructure of an air-cooled carburized Fig. 5.31 Hardenability curves and corresponding
barend.50x microstructures for a 25CrMo4 steel. Source: Ref 52
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
118 I (arburizing: Microstructures and Properties

doubled by special treatment to remove the mi- tigue strength of the ESR steel is improved even
crosegregation. An electroslag melted steel was further by applying a special treatment to reduce
also improved by the same balanced mechanical the severity of the microsegregation. The fatigue
and thermal treatment. At tensile strengths typi- limit of the conventionally melted steel is little
cal for the core of a carburized part made from a enhanced by the special treatment, although in
wrought blank, it is unlikely that microsegrega- the high-stress, low-cycle region, the treatment
tion has much influence on ductility and tough- did lead to an improvement.
ness indicators in the longitudinal direction. On What these tests suggest is that for high-cycle
the other hand, it does have a pronounced nega- fatigue applications, the effect of nonmetallic in-
tive affect on those properties in the transverse clusions far outweigh any adverse influences of
direction (Fig. 5.32). microsegregation. Presumably, the nonmetallics
Influence on Fatigue. It is difficult to deter- provide sites at which fatigue cracks can initiate
mine the effect of microsegregation alone on and from which they can grow. In the low-cycle
the various properties of a steel because of the domain (in this example, <2 x lOS cycles), micro-
presence of nonmetallic inclusions. Similarly, it segregation appears to be more influential than
is difficult to isolate the influence of nonmetal- nonmetallic inclusions.
lic inclusions on properties. However, if it is as- These test results (Ref 53,54) refer to a high-
sumed that electroslag remelted (ESR) steels are strength, through-hardened material, and
essentially cleaner than the conventionally whether the same trends apply to a carburized
melted steels, then some measure of the respec- and hardened part is not really known. Nonethe-
tive influences of microsegregation and nonme- less, some speculation can be made. For exam-
tallic inclusions can be obtained. ple, for the core, there will be a similar trend:
Kroneis et al. (Ref 53, 54) carried out such a microsegregation can affect low-cycle fatigue
comparison using a Cr-Mo-V steel and showed and have little, if any, influence on the high-
the ESR steel to have superior general properties. cycle fatigue strength. For the case, the local trans-
In this way, the effect of nonmetallic inclusions formation behavior might be affected by micro-
in particular could be assessed. With a special segregation. However, the effect will be detri-
treatment to reduce the microsegregation of the mental only if it leads to the formation of mixed
steels melted by the two processes (ESR and microstructures, such as martensite plus bainite
conventional), the properties of each were further or martensite plus austenite, depending on the
improved. For example, the bending fatigue grade of steel. If, for example, the retained aus-
strength of the ESR material (Fig. 5.33) was su- tenite at a carburized and hardened surface is
perior to that of the conventionally melted steel. patchy, varying appreciably over short distances
Therefore, the conclusion is that nonmetallic in- due to microsegregation (Fig. 5.34a, b), then the
clusions have a significant influence on the fa- microhardness will vary accordingly, as too will
tigue strength of this particular high-strength, the local residual macro- and microstresses. The
through-hardened, and tempered steel. The fa- bending fatigue strength would then approach
that of the softer transformation product.

36 8!. ±900 . - - - -......-....,....~==-==-===-,
<f1. 32
\ 5;eel: ~38 CrMoV 51 ~

..; 28 \ after four-fold reduction £Cl ±800 1--.....:::~----:~~F=,=,;==::~:7=::-+

~ 24 ~
.~ 20 :\. 'Iii ±700
:g 16
\ ~

I, g ±SOO 1----:1----+----+----::1
~ .l!!
~ 12 Cl
8 r-... =6 ±500 1----"""'oE''''''''"=y.
""1'0 ~ :"''C'''''''''-'f''=';=''''''iF'
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
Degreeof segregation: (MomaxokIMomin %)

Fig. 5.32 Dependence of reduction of area of trans-

verse specimens on degree of molybdenum segregation.
Hardened and tempered; tensile strength, 1550 to 1770
MPa. Source: Ref 54
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 119

Other Aspects of Microsegregation. A danger that give rise to tearing through the ferrite and its
period for heavily microsegregated steels, espe- adhesion to the tool tip.
cially the more alloyed steels, is during a slow
cool after carburizing. The slow rate of cooling
can lead to a carburized layer microstructure that Nonmetallic Inclusions
consists of distinct areas of HTIP residing
alongside areas of low-temperature transforma- Origin of Nonmetallic Inclusions
tion products. The short range residual stresses
developed during the cool can be very high and Steels. are not commercially made free from
can induce cracking. nonmetallic inclusions, nor are they likely to be.
Another danger period is during the grinding It is estimated that there are millions of inclu-
of the carburized and hardened surface. If the sions per tonne of steel. Table 5.7 indicates this
structure consists of alternating areas of austenite estimate for a hypothetical steel in which the ox-
and martensite, each of these features will react ygen and the sulfur contents are 1 ppm. In actu-
differently to the action of the abrasive wheel. ality, steels contain more oxygen and sulfur (> 10
Grinding will deform and possibly tear the aus- ppm); however, increasing the oxygen and sulfur
tenite volumes in the surface, and also cause the do not so much increase the number of inclu-
immediate surface temperature to rise more than sions, but rather is more likely to increase their
would a wholly martensitic surface. Therefore, average size. Also, inclusions within a piece of
burning and cracking will tend to occur more steel are not one size, or close to one size, but are
easily in a surface that contains alternating areas within a wide range of sizes. A clean steel, there-
of martensite and austenite (See Chapter 8 on fore, may have 1012 inclusions per tonne where
grinding). most are less than 0.2 urn in diameter and few
With regard to machining during the early exceed 20 urn. An average steel has more inclu-
stages of manufacture, microsegregation can be sions of this size range, and dirty steels contain
responsible for heavy tool wear, although much many inclusions having sizes much larger than
depends on the actual microstructure. Banded 20~m.
structures, consisting of layers of ferrite and In steels, there are two classes of nonmetallic
pearlite, each one grain thick, machine better inclusions: exogenous and indigenous. The for-
than those where bands are several grains thick mer occurs as a result of pieces or particles of

Fig. 5.34(a)
Flow of microsegregation caused by Fig. 5.34(b) Nonmetallic inclusions and banding in a
forging in a Ni-Mo steel part. Note the surface carburized heavily microsegregated 1 % C alloy steel sample. Re-
layer. tained austenite, light
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
120 / Carburizing: Microstructures and Properties

refractory material being separated from surfaces enced metallographer, a study of the inclusions
of the steelmaking equipment with which the observed within a sample of steel can be of value
molten steel has made contact (e.g., furnace lin- in assessing how that steel was made.
ings, runners, and ladles). Such inclusions can The distribution and the average size of non-
range from extremely small to quite large lumps. metallic inclusions in a finished ingot is not usu-
They can provide surfaces onto which the indig- ally uniform. It depends on the melting and the
enous types of inclusions can nucleate, grow, casting-pit practices and the method and materi-
and with which they can chemically react. The als of the deoxidation sequence. For example, in
indigenous inclusions form as a result of reac- a killed steel, the largest population of oxide in-
tions that take place during steelmaking and the clusions is in the bottom part of the ingot. In-
deoxidation processes (deoxidation products). creased aluminum additions progressively re-
Primary deoxidation takes place in the melt at duce that population. Sulfides tend to be smaller
high temperatures, followed by secondary in the columnar solidification zone than in the
deoxidation during the liquid cool, and to some equiaxial solidification zone, and the central
extent when the steel has solidified. Much of the zone of the middle of the ingot contains more
nonmetallic material present in deoxidized liquid sulfide than either the top or the bottom of the in-
steel rises to the surface, leaving a fair quantity got.
either trapped in the dendrites growing in from Anisotropy and Nonmetallic Indusion Shape
the surface or carried to the bottom of the ingot Control. One problem encountered by steel us-
by falling dendrites. Small inclusions, which rise ers is the difference between longitudinal and
only slowly in the liquid metal, are more likely to transverse properties, called the anisotropy effect
be retained in the steel. (Table 5.8). This effect is primarily related to
The products of deoxidation are the oxide the presence and behavior of deformable non-
types of inclusion, and the oxides of most impor- metallic inclusions and, to some extent, micro-
tance include aluminum oxide, Al203, and sili- segregation, which become elongated in the di-
con dioxide, Si02 . The main product of the rection of working. Manganese sulfide is a sig-
desulfurizing reactions is manganese sulfide, nificant contributor to the anisotropy effect;
MnS. Without adequate quantities of manganese it is soft and deformable under cold and hot
present in the steel (>0.3%), iron sulfides form, working conditions. However, the ability to de-
leading to problems of hot-shortness during form tends to increase with type (I, Il, or ill)
metalworking operations. These compounds are and decrease with substitution of manganese by
the pure forms, but they can and do combine other elements in the system, such as iron,
with other inclusion-making materials, such as chromium, and nickel, which raise the hardness
MgO·Al203 or iron and manganese silicates. of the inclusion (Table 5.9). Calcium aluminate
The oxide, Si02 itself can have modifications (CaO·AI203) and corundum (Al203) do not de-
(cristobalite, tridyrnite, and quartz). The manga- form, and pure silica inclusions are brittle at met-
nese in MnS can be substituted in varying de- alworking temperatures. Therefore, unless they
grees by other elements present in the system, are present as stringers, they contribute less than
such as iron and chromium. Furthermore, inclu- MnS to the anisotropy effect. Iron-manganese
sions of one type can be found attached to or silicates are deformable above 850°C, so in
within inclusions of another type. Thus, the sub- hot-worked steels they too contribute to the dif-
ject of nonmetallic composition and identifica- ference between longitudinal and transverse
tion is quite complex. Nevertheless, to an experi- properties.

Table5.7 Influence of indusion size on indusion numbers in a hypothetical steel with 1 ppm oxygen and
1 ppm sulfur as spherical Alp] and MnSindusions
IDclusion diamet.! f!Dl No.oflnclusions(l Volume of steel per Inclusion Mean distanee between Inclusions
103 2.3 X UP 55cm 3 3.8cm
102 2.3 X 106 55mm 3 3.8mm
10 2.3 X 109 55 X 106~m3 380~m
1 2.3 X 10 12 55x UP~m3 38~m
10-1 2.3 X 10 15 55~3 3.8~m
10- 2 2.3 X 10 18 55x 1O-3~3 3.8~m
lOA 2.3 X 1021 55 X IO...{;~3 380A

Source:Ref 55
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 121

Steelmaking processes, such as electroslag component is replaced by manganese from the

remelting (ESR) or vacuum arc remelting steel matrix, thereby producing a truer MnS par-
(VAR), reduce the amount of significantly sized ticle surrounded by manganese depleted steel.
nonmetallic materials within a steel (Ref 57). High-temperature heat treatments can also bring
The ESR process produces greater freedom from about shape changes (e.g., rods of type II MnS
sulfides, whereas VAR gives the steel a lower ox- can spheroidize) (Ref 59), composition and
ide content (Ref 55). Electron beam remelting phase changes (Ref 60,61), and the precipitation
appears to effectively reduce both sulfides and of one type of inclusion within another (e.g., CrS
oxides (Ref 58). within MnS). In some instances, partial or com-
Calcium, rare earth (RE), or titanium treat- plete dissolution of sulfides can occur (Ref 62),
ments further decrease the number of visible in- followed by reprecipitation at prior austenite
clusions and better distribute and modify them. grain boundaries, as in overheating.
These treatments make the manganese sulfide in-
clusions less deformable during metalworking Effects of Nonmetallic Inclusions
processes, thereby significantly reducing the ani- Influence on Tensile Properties. Nonmetallic
sotropy effect (Ref 58). A reduced anisotropy is inclusions, typical of the commercial grades of
of special interest to plate users and fabricators. wrought steels, do not significantly influence the
The composition, size, and distribution of non- ultimate tensile strength or the yield related
metallic inclusions within a steel are initially de- properties of a material in either the longitudinal
termined by the steelmaking parameters, such as or transverse directions. On the other hand, the
the boiling time, the refractory type, the deoxi- ductility indicators (reduction of area and elon-
dizing materials, the deoxidation practice, and gation) are affected, at times appreciably so.
the tapping and teeming methods. Modern Thus, for two steels that are identical apart from
steelmaking methods, in particular the remelting their nonmetallic counts, the stress-strain curves
technology, coupled with a good understanding are basically the same up to the point of maxi-
of the compositions and properties of nonmetal- mum stress. Beyond the maximum stress, when
lic materials, make it possible for the steelmaker an unstable condition develops and necking oc-
to adjust the cleanness of a steel according to the curs, the cleaner steel necks down more (a larger
eventual application. reduction of area) and stretches further (larger
Stability. At one time it was thought that non- elongation) than the less clean steel (Fig. 5.35).
metallic inclusions were relatively stable. How- For a given steel, the reduction of area tends to
ever, because reactions in such materials take fall as the tensile strength increases. The effect,
place fairly slowly, inclusions are regarded as be- however, becomes more pronounced as the in-
ing initially metastable (in the as-cast ingot). For clusion count increases, as Fig. 5.36 (Ref 58)
example, after cooling, the sulfur within the in- shows (where values of true strain Er were deter-
got may be tied into an iron-manganese-sulfur mined from the cross-sectional areas at the frac-
inclusion where the iron content is high. How- tures and, therefore, reflect on the reduction of
ever, during subsequent heat treating, the iron

Table5.9 Hardness of nonmetallic inclusions

Table 5.8 Effectof sulfide inclusions on NoDJDetall1c: iDdusion Microbardness, Iq;f1DunZ
mechanical properties in a Cr-Ni-Mo steel Manganese sulfides(a)
Property Many Inclusions Few iDe_ions MnS 170
Ultimate tensile strength. MPa 775 (79) 700(72) MnS+Cr <450
(kg/mm 2) MnS+Ni <250
Reduction of area. % MnS+Co <240
Longitudinal 62.8 59.5 MnS+Fe <300
Transverse 17.5 29.7 MnS+Ti <215
Reduction of area anisotropy. 0.28 0.50 MnS+V <340
transverse/longitudinal AI20 3 >3000
Bending fatigue. MPa (kg/mm 2) Si02 1600
Longitudinal 390(39.4) 375(38.5) MnSi0 2 750
Transverse 320(32.4) 320(32.6) MnO 400
Fatigue anisotropy. transverse! 0.82 0.85
longitudinal Nonmetallic inclusions with MgO present. usually from refractories,
Fatigue ratio. longitudinal fatigue! 0.5 0.54 tend to be hard (1000 to 1200 kgf/mm 2) or very hard (2100 to 2400
ultimate tensile strength kgf/mm 2), such as MgO·AI203. (a) Room temperature. includes
manganese sulfides in which manganese has been substituted by
Source: Ref56 other elements. Source: Ref 55
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
122 / Carburizing: Microstructures and Properties

area). Ductility indicators are adversely affected

by any type of nonmetallic, but the stringer type
Maximum is more detrimental than the globular type (Ref
63). Also, the more elongated the stringer is, the
worse the effect will be.
The anisotropy effect due to metalworking has
already been mentioned, though not in any de-
tail. The ductility indicators in conventionally
produced steels are lowest in the short transverse
direction, intermediate in the long transverse di-

rection, and highest in the longitudinal direction
(Ref 64). The variation of reduction of area

(ROA) relative to orientation is illustrated in Fig.
5.37. An example of the effect that forging ratio
has on the ROA is shown in Fig. 5.38, which
shows that anisotropy increases with the amount
of metalworking. However, these results are
from about 1949; since then, steelmaking and
metalworking processes have advanced appre-
ciably so that differences due to orientation can
be better controlled (Table 5.10). In the cleaner
Extension grades of steel, the difference in ductility be-
Fig. 5.35 Load/extension curves indicating the trend tween the longitudinal and long transverse direc-
formed by steel cleanness tions can be kept small, although that for the

1.50 r-----,---...,--..,--,----...,------,---,--~_____,

- - 900 MPa tensile strength

- - - 1150 MPa tensile strength
1.25 0:--+---+-+--- 1850 MPa tensile strength


0.25 ~---+--l--+-+--'::~~W

0.04 0.06 0.080.10 0.20 0.40 0.60 0.80 1.0

Inclusions, mm/mm2

Fig. 5.36 Scatterbands for the tensile true fracture strain of a 0.4% C steel at three strength levels as a
function of the total inclusion projected length in the fracture plane. Inclusions, total projected length.
Source: Ref58
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 123

short transverse direction lags some way behind. tallic inclusion with respect to fatigue stressing
Even so, it might not be necessary to make steels depends on the chemistry, size, location, and
with equal ductility in all directions. The impor- quantity of inclusions; strength of the steel; and
tant achievement would be to have adequate duc- the residual stress state immediately adjacent to
tility in the short-transverse direction, and this the inclusion (Ref 66).
can be done by producing a low-sulfur steel Effect of Inclusion Chemistry. The hard types
along with a RE treatment. of inclusions that resist deformation during pro-
Both microsegregation and nonmetallic inclu- cesses, such as rolling or forging, have a ten-
sions vary depending on where in the forging or dency to develop cone-shaped cavities in the di-
bar the samples were taken. Consequently, be- rection of metal flow during the working
cause they both have an effect on the ductility in- operations (Ref 67). These cavities, which do not
dicators, the ductility should vary with location. weld up again, increase the effective size and the
Table 5.11 shows that near surface ductility is su- stress-concentrating effect of the inclusion,
perior to that at the mid-radius or center loca- thereby becoming more potent as a fatigue crack
tions. This comparison might be of interest to initiation site. These hard, undeformable types
gear manufacturers who, in general, cut the gear of inclusions, such as single-phase alumina
teeth in the near surface region of the forging or (Al z0 3) , spinels (e.g., MgO·Al z0 3), calcium
bar. aluminates (e.g., CaO-AIZ0 3), titanium nitride,
Influence on Fatigue Resistance. Leslie (in Ref and some silicates, all have lower thermal expan-
65) considered that in the high-cycle regime sion coefficients than steel (Fig. 5.39). There-
(> 1OS load cycles) nearly all fatigue cracks initi- fore, a tensile stress field should develop around
ate at nonmetallic inclusions. In the low-cycle re- each inclusion as a result of hardening heat
gime (l()3 to lOS load cycles) and especially in treatments, and the tensile stresses could then
the 101 to 1()3 cycle range, slip band cracks pro- add to any stresses applied in service. Thus, with
vide the initiation sites. The potency of a nonme-
Longitudinal Transverse 68

~ 64
~60 l!!

<, <, Maximum '0
as 60
<,r-- '0 56
\ :::>
II: 52
~ 30
au =1145 MPa \ Minimum

o ~ 40
60 60 100
Angle, degrees Forging ratio
Fig. 5.37 Relationship between reduction of area and Fig. 5.38 Effect of forging reduction on longitudinal
angle between the longitudinal direction in forging and and transverse reduction of area. TensiIe strength, 840
the specimen axis. Source: Ref 64 MPa (118 ksi), Source: Ref 64

Table5.10 Mechanical properties of 15NiCrMo 16 5 carburizing steel after melting (after K. Vetter)
Sulfur Degree of CUproof lenslle Elongation attracture(a), "" Reduction ofarea, % Notch toughness(h), J/cnJ1.
Meltllllltype content, % delbrmation stress, MPa stress,MPa Longitudinal 'Iransverse Longitudinal 'Iransverse Longitudinal 'Iransverse
Conventionally 0.010 -6x 1160 1450 10 7 48 27 50 35
Electroslag 0.004 -6x 1160 1450 12 10 55 46 69
Electron-beam 0.007-0.009 10 to 30x 1060--1120 1350-1420 13.5 13.5 58 51 83 65

(a) lu =5do. (b) DVM specimen. Source: Ref 57
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
124 / Carburizing: Microstructuresand Properties

Table 5.11 Mechanical propertiesof heat-treatedforged barsof X41 CrMoV 51 steel

Positioool Iagot type of EIongalion at fracture(a) Reductionofarea, " Radool~specimeo
specimen in ioRot -Jtinc Loqltudioal 'I'raonene LoogitudioaJ Transverse EIoogation Reductionofarea
Edge Conventionally melted 12 6 45 12 0.5 0.27
E1ectroslag remelted 12 12 45 43 \.0 0.97
Y, radius Conventionally melted 10 2 34 5 0.2 0.15
Electroslagremelted 10 10 40 38 \.0 0.95
Center Conventionally melted 10 2 30 4 0.2 0.13
Electroslagremelted 10 9.5 40 36 0.95 0.90

Diametersfrom 230to 350 mm;tensilestrengthsfrom 1550to 1700MPa;averagesof -200 heats.(a) 4, =5do. Source:Ref 57

respect to properties, the hardest nonmetallic in- veloped by manganese sulfide. This is due in part
clusions are thought to be the most harmful (Ref to the deformability of manganese sulfide and in
68-70), and the softer inclusions, such as manga- part to the tendency of hard nonmetallic inclu-
nese sulfide and manganese oxide, are the least sions to be noncoherent with the ferrous matrix,
harmful (Ref 70). The softer inclusions are es- whereas manganese sulfide is semi- coherent.
sentially harmless even in high-strength bearing In general, the term "nonmetallic inclusion"
steels (Ref 60), although there could be a little refers to the exogenous and indigenous types.
more life variability. A hard inclusion can be However, a third type develops by the inten-
rendered less deleterious if it contains or is en- tional addition of lead, which is essentially im-

Mean expansion coefficient (0-800 ec), U1X 10-f3rC

5 10 u2 15

c = c ao )
( A=AI


(XOAI 20 3)
~c. Nitrides
o Silicates (a)
(3 c=cao)
.!: A= AJ203
13 ( M=MnO
a S=Si02


Stress-raising potential Matrix

? Void forming potenti~1

Fig. 5.39 Stress-raising propertiesof inclusions in 1%C-Crbearing steel.

Source: Ref55
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features /125

miscible in steels and which fOnDS a fine dis- Effect of Inclusion Quantity. For an inclusion
persion within a lead-treated steel. It is added to initiate a fatigue crack, it must be in the path
for the sole purpose of giving the steel free- of the applied stress. Often that path is quite nar-
machining properties. Lead is soft and, there- row, such as at the fillet of a gear tooth. The
fore, contributes to anisotropy in wrought steels probability of an adequately sized inclusion be-
similar to manganese sulfide. Goldstein et al. ing critically located within a narrow load path
(Ref 71) found that lead reduced the fatigue increases as the number of adequately sized non-
limit by about 8% (compared with unleaded metallic inclusions in the steel increases. Figure
steels) but anisotropy accounted for an -30% 5.41 shows how the fatigue limit falls as the in-
difference in the fatigue limits of both leaded clusion count increases. Crack propagation can
and unleaded steels. be assisted or impeded by nonmetallic inclusion,
Effect of Inclusion Size and Location. Critical perhaps depending on orientation.
inclusion size varies depending on the composi- In design, it is not unusual for a stress concen-
tion of the inclusion, because composition deter- trator to be added close to another stress concen-
mines shape (round or angular) and whether or trator in order to reduce the stress raising effect
not residual stresses, or cavities, are associated of the original. Therefore, with respect to non-
with the inclusion. Critical size increases with metallic inclusions, is it possible that neighbor-
distance from the surface; the further an inclusion ing inclusions could reduce the stress raising po-
is from the surface, the larger is its critical size. tency of each other? In other words, are two 40
For alumina inclusions, the critical size for an in- urn inclusions that are close to one another less
clusion close to the surface is 10 urn, but it in- harmful than an isolated inclusion of the same
creases to 30 urn when the inclusion is located diameter?
100 /lID from the surface (Ref 72). Kawada et al. Effect of Steel Strength. For a critical location
(Ref 73), studying the bending fatigue strength of within a steel, the critical inclusion size is
hardened bearing steels, agreed with the 10 urn smaller the harder the steel is (Fig. 5.42). In
critical defect size for a surface inclusion, but they high-hardness steels including case-hardening
were not so optimistic regarding the critical size at steels, calcium aluminates are the most detri-
100 /lID (Fig. 5.40) unless the local data are aver- mental of inclusions (Fig. 5.43). Calcium alumi-
aged. When a subsurface inclusion is above the nates are far more harmful than, for example,
critical size, its strength reducing effect is propor- alumina or titanium carbonitride, shown in Fig.
tional to the cube root of its diameter (Ref 72). 5.44 (Ref 65). Manganese sulfide is not particu-
The critical inclusion or defect size with re- larly damaging (Ref 56), especially in relatively
spect to fatigue loading is not the same for frac- low-strength materials (Table 5.8).
ture toughness or metalworking situations.

E 0.15 Tensile strength, MPa
.§ "
:::> os
~ 0.10 a.
:2 800
o C
!E 0.05 .. .," "
0 ic1
I/o ••••
.... ;.. ............. Q)
Ol 600
81 ••I/o •••
~ 0 Ol
.c:., 500
o 0.2 0.4 0.6 0.8 1.0 1.2 -E
Distance from surfaceto centerof inclusion, mm ::J

Composition, ...
Steel C SI Mn P S Cr Ni Cu
0 1.04 0.27 0.40 0.013 0.011 1.43 0.08 0.15 300

0.020 0.020- 1.65
0.012 0.006 1.39 0.17 0.14
10- 2 10-1
Inclusions, mm/mm2

Fi~. 5.40 Fatigue fracture of bearing steel caused by Fig.5.41 Relationship between the limiting fatigue
inclusions in rotary bending. Distance from steel surface stressin smooth, rotating bending tests and total inclusion
vs. diameter of inclusions that initiated fatigue fracture. projected length. Results for two test orientations and
Source: Ref 60,73 three strength levels are shown. Source: Ref 58
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
126 / Carburizing: Microstructures and Properties

Oa at 10
reversals = ~ _

Mean inclusion radius (rl. in.

Fig. 5.42 Effect of inclusion size (spherical) on the fatigue
strength of steel. Source: Ref 74

Effect ofResidual Stresses from Inclusions. As which implies that the critical inclusion size is
previously mentioned, tensile stress fields can small and that other adverse features are also
develop around the harder type of nonmetallic present. However, a case-hardened surface with
inclusion, and cavities can form at hard inclu- its compressive residual stresses can better toler-
sions during metalworking operations to act as ate such adverse features than can a residual
stress raisers. Also, angular inclusions (Ref 75) stress-free surface.
or strings of fragmented inclusions can act as Any surface treatment that induces harden-
stress raisers, even without the presence of cavi- ing and compressive residual stresses offsets
ties. In through-hardened and low-temperature much, if not all, of the adverse influence of
tempered steels for which the inclusion critical nonmetallic inclusions. Cook and Dulieu (Ref
size is small anyway, such features are detrimen- 58) and others have demonstrated this with
tal to the fatigue resistance of the steel. In case- shot peening. Having said that, Fig. 5.43 shows
hardened parts, the surface layers are hard, the damaging influence of calcium alurninates

1000 9O.-------,---:-r---r----r---,

Not SiCa injected /
800 50

700 /
/. ~ 'if!.

~// / SiCa injected

500 lV
;, V

200 300 400 500 600 700 800 106

Hardness, HV Life, stresscycles
Fi~. 5.43 Fatigue limits for SiCa-injected and not Fig. 5.44 Rotating bending fatigue of samplesinitiated
SiCa-injected steels at different hardness levels. The by B (alumina, irregular), D (calcium aluminate, spheri-
SiCa-injected steel with D type inclusions has a lower fa- cal), and T (titanium carbonitride, cuboid) type inclusions
tigue limit. Source: Ref 65 in an SAE 52100 steel. Source: Ref 65
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 127

on the fatigue limit of case-hardened test Influence on Contact Fatigue. As in other cy-
pieces, although it is not clear if the fractures clic loading situations, nonmetallic inclusions
were surface or core initiated. It is not uncom- can contribute to failure during rolling or rolling
mon, in studies related to the fatigue strength with sliding, such as in bearings and gears.
of case-hardened test pieces, for failures to ini- Again, size, hardness, coherency with the matrix,
tiate at inclusions within the case or in the core location, and the number of the inclusions are the
just beneath the case where the residual important factors.
stresses are tensile (Ref 76, 77). Such failures Hard inclusions situated between the working
occur in the high-cycle area of the S-N curve, surface and the depth of significant Hertzian
and they are attributed to the presence of hy- stresses are potentially damaging. The more in-
drogen concentrated at the site of an inclusion clusions there are above a critical size and resid-
(Ref 65). ing in the stressed layer, the more likely they are
to reduce the life of the part (Fig. 5.45). There-
fore, clean steels with inclusion size control
15 should be more resistant to failure.
0 The stress associated with a hard inclusion is
Medianfor 50 rings
tensile and at a maximum at the interface be-
tween the inclusion and the steel matrix. Ac-
cording to Winter et aI. (Ref 78), the stress-
'U 10 raising effect of an oxide inclusion is 2.5.
~ Therefore, some point on the interface, de-
pending on aspects of the inclusion and the di-
==CD rection of the applied stresses , is where dam-
E' age or cracking eventually initiates. If a cavity
~ 5 exists at an inclusion (due to forging) or a
crack develops due to loading , the stress con-
centrating effect of the inclusion increases. If,
on the other hand, shearing (or whatever mech-
OL-- ----l --L _
anism is involved) occurs, which gives rise to
50 100 150 200
white etching areas associated with the inclu-
Oxide inclusionslarger than 0.03 mm
sion called a "butterfly" (Fig. 5.46), then the
stress-raising effect of the inclusion is reduced
Fig. 5.45 Effect of number of large ox ide incl usions on
the flaking of rig-tested bearings (1309 outer rings). (Ref 79).
Through-hardened steel for ball bearings; composition, Soft inclusions are not regarded as damaging
1C, 0.5Mn, 1.5Cr. Source: Ref 60 in rolling contact situations. Even so, they could

Fig. 5.46 W hite etchi ng "butterfly," developed at a nonmetall ic inclusion as a result of con tact loading. 750x
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
128 / Carburizing: Microstructures and Properties

have some effect if they lie in the zone of cyclic sive to the flank of a tool, and it provides a pro-
stress. Littman and Widner (Ref 79) state that in tective layer on the rake face, which protects
bearing fatigue tests, cracks and butterflies are against cratering.
almost never associated with sulfide inclusions. The crystalline silicates and alumina (the
Sugino et al. (Ref 80) confirm that MnS is fairly nondeformable types) tend to give heavy tool
harmless in this respect, but Al 20 3 (with or with- wear and are not effective as chip breakers.
out associated MnS) encourages the formation of Those silicates that become deformable at about
butterflies (Table 5.12). Winter et al. (Ref 78), 850°C assist machinability when the cutting
referring to gears, observed very small areas of speed is sufficiently high to generate tempera-
white etching material at sulfide inclusions but
tures in excess of 850 °C in the swarf metal as it
did not associate them with contact failures.
shears away from the rake face of the cutting
Consequently, soft inclusions are of less concern
than the harder types. Nevertheless, their size,
The ceramic and the coated tools developed
type, and quantities are best controlled to im-
for clean steel machining might not be durable
prove other properties.
for machining the less clean steels. Sulfide inclu-
Influence on Impact Fracture Strength and
sions have the potential to chemically react with
Toughness. The Charpy toughness (shelf energy)
the cutting surfaces of such materials, thereby in-
is increased by increasing the cleanness of a steel
(Ref 81) or by reducing the average size of inclu- creasing tool wear.
sions and the number of large inclusions. The Overheating. The appearance of matte facets
transition temperature, on the other hand, is little on the fracture faces of impact test pieces indi-
if at all affected by an increase of cleanness. cates that the steel has been "overheated" dur-
The toughness in the longitudinal direction is ing the preheating and hot-working processes.
not necessarily improved by increasing clean- The dissolution of manganese sulfide and its
ness, but for each of the transverse directions, es- reprecipitation as very fine particles on the aus-
pecially the through-thickness direction, tough- tenite grain boundaries during hot working are
ness is improved (Ref 58,81). regarded as the cause (Ref 58). Decreasing the
sulfur content lowers the onset temperature for
Consequences of Producing Clean Steels overheating where the onset temperature is at its
lowest when the sulfur content is between 0.002
Machinability. The improvements to metal and 0.004%. The steel composition and the
shaping machines and tools have not always
cooling rate from the working temperature have
matched the improvements in steel cleanness. As
an effect on overheating. For example, nickel
a result, machining problems are encountered
lowers the overheating temperature, and cooling
when the sulfur content of a lean-alloy steel falls
below about 0.009%. Difficult machining can rates between 10 and 300 °C/min seem to pro-
lead to poor workpiece surface finish and heavy duce more matte facets on the impact test frac-
tool wear. Even when the sulfur content is above ture surfaces than other cooling rates (Ref 82).
0.009%, the machinability can vary somewhat A reheat temperature of 1150 °C is considered
depending on the size of the sulfide particles. safe, and in the event of overheating, a reheat to
Steels with larger particles tend to machine more 1200 °C reduces, if not eliminates, the effect.
easily than do those in which the sulfides are Thus, overheating is reversible. According to
small. Ref 59, RE or calcium modifications to a low
Manganese sulfide is well known for its con- sulfur steel can prevent overheating by raising
tributions to easy machining and good tool life. the overheating temperature.
It assists in chip breaking of the swarf as it flows McBride (Ref 82) asserts that severe overheat-
from the cutting zone. It is not particularly abra- ing can reduce the fatigue resistance by 20 to

Table5.12 Listof inclusion type and occurrence associated with butterfly formation
'IYpe of IIOIIIIIeIlIIIk indus;oo Appearance orooometallic inclusion Frequency or butterflies
MnS Elongated (<3 11m) Rare
Elongated «311m) Few
Verythin and long None
Finely dispersed stringers Many
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 129

25%. Gardiner (Ref 83) determined the best steels, a low-hydrogen welding practice is rec-
combination of preheating temperature and ommended.
percent metalworking reduction to maximize
the overheating effect of two VAR S82 case-
hardening steels. Using these conditions to in-
duce overheating, test pieces were prepared (car- Summary
burized and uncarburized) and tested. Over-
heating had no detrimental effect on: Grain size
• Rotating bending fatigue strength of either Grain size affects both case and core proper-
carburized and uncarburized samples ties; small grain sizes are required.
• Rate of crack growth in uncarburized samples
• Fatigue crack initiation sites • Preprocess considerations: Purchase steel of
• Tensile strength appropriate quality; adhere to the grain-size:
• Fracture toughness requirement. A normalizing heat treatment fa-
vors an initially uniform and small grain size.
It did, however, reduce the tensile ductility, the im- • In-process considerations: High-temperature
pact toughness, and the K ISCC• carburizing (> 1000 "C) tends to encourage
Burning takes place at temperatures above some grain growth. Direct-quenched cases
-1450 °C and is irreversible. It is caused by tend to be coarse in terms of grain size,
liquation at grain boundaries, encouraged by the martensite plate size, and the size of austenite
segregation to the boundaries of solute ele- volumes. Re-heat quenching favors a refined
ments, such as phosphorus and sulfur. When the structure.
sulfur content is high, the temperature for burn- • Postprocess considerations: Consider reheat
ing is not much above that for overheating. quenching, keeping in mind the distortion as-
When the sulfur content is low, the overheating pect. For some applications, shot peening
temperature is lowered, thereby increasing the might counter some of the adverse effects of a
range between the overheating and burning coarse-grained structure (although no data are
temperatures. available to show this).
Welding. It is not good practice to weld parts • Effect on properties: A fine grain size can oc-
before or after carburizing and hardening. How- casionally have an adverse influence on the
ever, because cleanness affects the response of a case and the core hardenability of lean-alloy
steel to welding, the following information is in- carburizing steels. However, in general, a fine
cluded. grain size is preferred. Coarse-grained steels
The use of cleaner steels significantly im- are thought to distort more during heat treat-
proves weldability. The risk of heat affected zone ment. A coarse austenite grain will, when
(HAZ) liquation cracking diminishes as the sul- quenched, produce large martensite plates and
fur content is reduced, and lamellar tearing be- large austenite volumes. Large martensite
comes less of a problem as the through- thick- plates are more prone to microcracks than are
ness ductility is improved as a result of steel small plates. Altogether, these adverse effects
cleanness. contribute to reduced fatigue and impact
However, clean steels are prone to HAZ crack- strengths.
ing during and following the welding operation. • Standards: Specifications on grain size call
This cracking occurs due to a high HAZ for a grain size of predominantly five or finer.
hardenability, which results from the relative ab-
sence of sulfide and oxide inclusions. These in- Microcracks
clusions, had they been present in quantity,
Microcracks refer to cracks that develop dur-
would have stimulated the formation of HTIP
ing the formation of plate martensite. They are
(Ref 84). Instead, there is a greater tendency for
confined to the high-carbon regions of the case.
martensite to form in the HAZ, which, coupled
Large high-carbon martensite plates are more
with the intake of hydrogen into the zone (from
likely to have microcracks.
the steel and from the welding), can result in hy-
drogen-induced HAZ cracking. Hewitt (Ref 85) • Preprocess considerations: Plain-carbon and
reported a definite trend between hydrogen very lean-alloy grades of steel are more prone
cracking and sulfur content. Therefore, for clean to microcracking during case hardening.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
130 / Carburizing: Microstructures and Properties

Therefore, the steel grade needs to be consid- • Effect on properties: See comments for "In-
ered, as does grain size. process factors" in this list. Microsegregation,
• In-process considerations: High carburizing along with nonmetallic inclusions, apprecia-
temperatures favor grain growth and, hence, bly influences the transverse properties of
coarse plate martensite. Microcracking ten- wrought steels, particularly the toughness and
dency increases with carbon content; there- ductility indicators. Austenite/martensite or
fore, consider case carbon control. Direct martensitelbainite banding, if either occur, is
quenching retains more carbon in solution in expected to have a negative influence on fa-
the martensite; therefore quenching method tigue resistance, especially in those surfaces
and temperature are important. containing bands of bainite.
• Postprocess considerations: Tempering is re- • Standards: No specifications
garded as beneficial, and refrigeration might
have a negative effect. Nonmetallic Inclusions
• Effecton properties: This is difficult to assess,
though it is thought by some that fatigue All steels contain numerous nonmetallic inclu-
strength is reduced by as much as 20%. sions, but the cleaner grades have fewer large or
• Standards: ANSII AGMA has no specification significant inclusions than do the conventional
for grades 1 and 2. For the grade 3 quality, 10 grades. Clean steels, however, are more prone to
microcracks in a 0.0001 in. 2 (0.06 mm-) field overheating problems during forging.
at 400 is specified.
• Preprocess considerations: Steel quality re-
quirement, and the steel purchased, is deter-
Microsegregation mined by the product quality sought, for ex-
ample, ANSIIAGMA 2001 grade 1,2,3, or
Microsegregation is unavoidable; it is influ- aircraft quality. For special applications, ul-
enced by the modes of solidification during trasonic surveys maybe carried out to deter-
casting. The degree of microsegregation is af-
mine if, and where, significant inclusions or
fected by the alloy content of the steel, including
inclusion clusters reside within the work-
tramp elements. The distribution and direc-
piece. Actions are taken accordingly. Clean
tionality of microsegregation are affected by
steels can be more difficult to machine than
metalworking processes, and its intensity can be
conventionally melted steels.
reduced to some degree during the mechanical
• In-process considerations: No in-process consid-
and thermal treatments. Generally speaking,
manufacturers and heat treaters have to live with
• Postprocess considerations: Shot peening of
critically stressed areas will, to some extent,
• Preprocess considerations: Microsegregation offset any adverse effects of nonmetallics re-
is difficult to assess during acceptance testing; siding at, or close to, the surface of that area.
it is often masked by interstitial elements, for • Effect on properties: Nonmetallic inclusions
example, by carbon in normalized products. It can have a significant adverse influence on
can create machining problems. transverse toughness and ductility. In terms of
• In-process considerations: Microsegregation fatigue properties, soft inclusions are less
can cause severe cracking in the more alloyed harmful than are hard inclusions, size for size.
carburizing grades during slow cooling from In case-hardened parts nonmetallics might not
the carburizing temperature. In carburized seriously affect the fatigue limit but could
and quenched surfaces, microsegregation can contribute to greater life variability at stresses
cause alternating zones of martensite and aus- above the fatigue limit.
tenite, which are regarded as unsatisfactory. It • Standards: For AGMA grade 2 and grade 3
can also influence the growth and distortion qualities, (ultrasonic and magnetic particle
behavior of a heat-treated part. nondestructive testing) requirements are spec-
• Postprocess considerations: Where actual case- ified. Sulfur contents to 0.040% are permitted
hardened parts are not sectioned for quality for grade 2, and to 0.015%S for grade 3.
assessment, the presence of microsegregation • Note: Magnetic particle examination may be
and its effect on structure will go undetected. needed for detecting defects at or close to a
Its presence should be considered during ser- steel surface; such defects are the more likely
vice failure analyses. to affect service life. In magnetic-particle ex-
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features/ 131

amination of gear teeth larger than 10 pitch Hart.-Tech. Mitt., Vol 21 (No.3), 1966,
normal diameter (p 00)' indicationsof less than 1~198
Y64 in. (-0.4 rom) are ignored, yet the critical 13.H. Dietrich, S. Engineer, and V. Schuller, Influ-
defect size for case-hardened surfaces could ences on the Austenitic Grain Size of Case
be less than this. Therefore, while such in- Hardening Steels, Thyssen Edelstahl Tech.
spections are important, they may not guaran- Ber., Vol 10 (No. I), July 1984, P 3-13
tee a freedom from harmful surface breaking 14. W.T. Cook, The Effect of Aluminum Treat-
inclusions. menton the Case HardeningResponseof Plain
CarbonSteels,Heat Treat. Met., 1984,(No. 1),
15.V. Franciaet al., Effectof the PrincipalMetal-
Grain Size lurgicalParameters of Steelon Carburising and
1. Nickel Low-Alloy Steels Case Hardening, In- Hardening, Metall. Ital., Vol74 (No. 12), Dec
ternational Nickel Ltd., 1968 1982,P 851-857
2. MA Krishtal and S.N. Tsepov, Steel Pr0p- 16.P.M. Kelly and J. Nutting, Strengthening
erties after High Temperature Vacuum Carbu- Mechanisms in Martensite, MG/Conf/85/65, p
rising, Metalloved. Term. Obrab. Met., Vol 6, 166-170
June 1980,p 2-7 17.MA Zaecone, J.B. Kelley, and G. Krauss,
3. M.l Vinograd, lY. Ul'Yanina, and G.A. Strain Hardening and Fatigue of Simulated
Faivilevich, Mechanism of Austenite Grain Case Microstructures in Carburised Steels,
Growth in Structural Steel, Met. Sci. Heat Conf Proc. Processing and Performance
Treat., Vol 17 (No. 1-2), JanlFeb 1975 (Lakewood, CO),ASMInternational, July 1989
4. B.L. Biggs, Austenite Grain-Size Control of 18.K. Onel and J. Nutting, Structure-Properties
Medium-Carbon and Carburising Steels, J. Relationship in Quenched and Tempered Car-
Iron Steel Inst., Vol 192 (No.4), Aug 1959,P bon Steel, Met. SeL, Oct 1979,P 573-579
361-377 19.RWK. Honeycombe, Steels-Microstruc-
5. 0.0. Miller, Influence of Austenitizing Time tures and Properties, Edward Arnold, 1982
and Temperautre on Austenite Grain Size of 20. P.Schane, Effectsof McQuaid-Ehn Grain-Size
Steel, Trans. ASM, 1951,Vol 43, P 260-289 on the Structure and Properties of Steel,
6. MA Grossman, Grain Size in Metals, with Grain-Size Symp., American Society for
SpecialReferenceto Grain Growthin Austen- Metals, 1934,p 1038-1050
ite, Grain-Size Symp., American Society for 21. E. Macherauch, Das Ermudungsverhalten
Metals, 1934,p 861-878
Metallischer Werkstoffe, Hart.-Tech. Mitt., Vol
7. RG. Dressel, R Kohimann, KJ. Kremer, and
22 (No.2), 1967,P 103-115
A. Stanz, The Influence of Rolling and Heat
22. V.D. Kat'ner, V.F. Nikonov, and SA Yurasov,
Treatment Conditions on Austenitic Grain
Modem Carburizing and Carbonitriding Tech-
Growth in Case Hardened Steels, Hart-Tech.
Mitt., Vol 39 (No.3), May/June 1984, p niques, Met. Sci. Heat Treat., Vol 15 (No.9),
112-119 Sept 1973,p 752-755
8. VA Kukareko, The Regularities of Austenite 23. J.L. Pacheco and G. Krauss, Microstructure
GrainGrowth in Steel 18KhNVA, Metalloved. and High Bending Fatigue Strength in Carbu-
Term. Obrab. Met., (No.9), Sept 1981,p 15-17 rised Steel, J. Heat Treat; Vol 7 (No.2), 1989
9. H. Child, Vacuum Carburising, Heat Treat. 24. H. Brugger, Werkstoff und Warmebehand-
Met., Vol3, 1976,p 60-64 lungseinflusse auf die Zahnfubtragfahigkeit,
10.J.D. Murray, D.T. Llewellyn, and B. Butler, VDI Berichte, (No. 195), 1973,P 135-144
Development of Case Carburizing C-Mn 25. H.-J. Kim and Y.-G. Kweon, High-Cycle Fa-
Steels, Low-Alloy Steels, lSI publication 114, tigue Behavior of Case-Carburized Medium-
The Iron and Steel Institute, 1968,p 37-53 Carbon Cr-Mo Steel, Metall. Mater. Trans. A,
11.L.N. Benito and D. Mahrus, 35thAnnual Con- Vol 27, 1996,P 2557-2564
gress ofAssociation Brasileira de Metals, Vol 26. G. Krauss,The Microstructure and Fractureof
1, Association Brasileirade Metals,July 1980 a Carburised Steel, Metall. Trans. A, Vol 9,
12.A. Randak and E. Kiderle, The Carburisation Nov 1978,P 1527-1535
of Case-Hardening Steels in the Temperature 27. GZ. Wang and J.H. Chen, A Comparison of
Range between 92~1100 °C (BISI 5231), Fracture Behaviour of Low Alloy Steel with
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
132 /Carburburizing: Microstructures and Properties

Different Sizes of Carbide Particles, Metall. Metall. Trans., Vol 5, Mar 1974, p 778-781
Mater. Trans. A, Vol 27, July 1996 42. T.A. Balliett and G. Krauss, The Effect of the
28. K. Isokawa and K. Namiki, Effect of Alloying First and Second Stages of Tempering on
Elements on the Toughness of Carburised Microcracking in Martensite of an Fe-1.22C
Steels, DenkiSeiko(Electr. Fum. Steel), Vol57 Alloy, Metall. Trans. A, Vol 7, Jan 1976, p
(No. I), Jan 1986, p 4-12 81-86
29.1.1. Prokof'eva, M.V. Taratorina, and LV. 43. C.A Apple and G. Krauss, Microcracking and
Shennazan, Effect of Martensite Needle Size Fatigue in a Carburized Steel, Metall. Trans.,
on the Mechanical Properties of Steel 45 after May 1973, Vol 4, p 1195-1200
Induction Hardening, Metalloved. Term. 44. M.A. Panhans and RA Fournelle, High Cycle
Obrab. Met., (No. 10), ocr 1978, P 10-13 Fatigue Resistance of AISI 9310 Carburised
Steel with Different Levels of Surface Re-
Microcracking tained Austenite and Surface Residual Stress,
30. RP. Brobst and G. Krauss, The Effect of Aus- 1. HeatTreat., VoI2(No.1),June 1981,p54-61
tenite Grain Size on Microcracking in 45. R.F. Kern, Carburize Hardening: Direct
Martensite of an Fe-1.22C Alloy, Metall. Quench vs. Reheat and Quench, J. HeatTreat.,
Trans., Vol 5, Feb 1974, p 457-462 March 1988, p 17-21
31. A.R. Marder and A.D. Benscoter, 46. 1. Lyman, High Carbon Steel Microcracking
Microcracking in Fe-C Acicular Martensite, Control during Hardening, J. Eng. Mater.
Trans. ASM, Vol 61, 1968, P 293-299 Techrwl. (Trans. ASME), Vol 106, July 1984, p
32. AH. Rauch and WR Thurtle, Microcracks in 253-256
Case Hardened Steel, Met. Prog., Vol 69 (No. 47. I. Laren and H. Fredriksson, Relations between
4), April 1956, P 73-78 Ingot Size and Microsegregation, Scand. J.
33. M.G. Mendiratta, 1. Sasser, and G. Krauss, Ef- Metall., Vol 1, 1972, p 59--68
fect of Dissoved Carbon on Microcracking in
Martensite of an Fe-1.39C Alloy, Metall.
Trans., Vo13, Jan 1972, p 351-353 48. T.B. Smith, Microsegregation in Low-Alloy
34. RG. Davies and C.L. McGee, Microcracking Steels,/ronSteel, Vol37,Nov 1964,p536-541
in Ferrous Martensites, Metall. Trans., Vol 3, 49.1. Philibert, E. Weinry, and M. Ancey, A Quan-
Jan 1972, p 307-313 titative Study of Dendritic Segregation in
35. RF. Kern, Controlling Carburising for Top Iron-Base Alloys with the Electron-Probe
Quality Gears, GearTechnol., MarlApril 1993, Microanalyser, Metallurgia, Vol 72 (No. 422),
p 16-21 Nov 1965, P 203-211
36. L. Jena and P. Heich, Microcracks in Carbu- 50. T.B. Smith, 1.S. Thomas, and R. Goodall,
rised and Hardened Steel, Metall. Trans., Feb Banding in a 1Y2% Nickel-Chromium-
1972, Vol 3, p 588-590 Molybdenum Steel, 1. Iron SteelInst., Vol 201,
37. AR Marder, AD. Benscoter, and G. Krauss, July 1963, P 602--609
Microcracking Sensitivity in Fe-C Plate 51. RG. Ward, Effect of Annealing on the Den-
Martensite, Metall. Trans., Vol 1, June 1970, p dritic Segregation of Manganese in Steel, J.
1545-1549 IronSteelInst., Vol203, Sept 1965, p 930-932
38. 1. Hewitt, The Study of Hydrogen in Low-Al- 52. A Kulrnburg and K. Swoboda, Influence of
loy Steels by Internal Friction Techniques, Hy- Micro-Segregations on the Transformation Be-
drogen in Steel, Special report 73, The Iron and haviour and Structural Constitution of Alloy
Steel Institute, 1962, P 83-89 Steels, Prakt. Metallogr., Vol6, 1%9, P 383-400
39. RG. Baker and F. Watkinson, The Effect of 53. M. Kroneis et al., BergHiittenmann. Montash.,
Hydrogen on the Welding of Low-Alloy Vol 113, 1968, p 416-425
Steels, Hydrogen in Steel, Special report 73, 54. M. Kroneis et. al., Properties ofLarge Forgings
The Iron and Steel Institute, 1962, p 123-132 of Electroslag Remelted Ingots, 5th Int.
4O.A.R. Marder and A.D. Benscoter, Forging Conj (6-9 May 1970),Temi,p 711-737
Microcracking in Plate Martensite of AISI
Nonmetallic Inclusions
52100 Steel, Metall. Trans., Vol 1, Nov 1970, P
3234-3237 55. R Kiessling, Nonmetallic Inclusions in Steel
41. A.R. Marder and A.D. Benscoter, IV, Metals Society, 1978; R Kiessling, Nonme-
Microcracking in Tempered Plate Martensite, tallic Inclusions in Steel II, Metals Society,
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Influential Microstructural Features / 133

1978 elusion Content, Tetsu-to- Hagane (J Iron

56. 1. Watanabe, Some Observations on the Effect Steel lnst.), Vol52, 11,Lecture229,103,1966
of Inclusions on the Fatigue Properties of 71. YE. Goldstein, A.Y. Zaslavskii, and B.S.
Steels, Proc. 3rd Japanese Congo of Testing Starokozhev, Effect of Lead on the Fatigue
Materials, Kyoto University, 1960,p 5-8 Characteristics of Structural Steels, Metal Sci.
57. E. Plockinger, Properties of Special Construc- Heat Treat., 5 May 1973,p 397
tional Steels Manufactured by Special 72. WE. Duckworth and E. Ineson, Clean Steels,
Melting Processes Including Steels for Special report 77, The Iron and Steel Institute,
Forging, Stahl Eisen, Vol 92 (No. 20), 1972, 1963,p 87-103
P 972-981 73. Y. Kawada, H. Nakazawa, and S. Kodama,
58. WT. Cook and D. Dulieu, Effect of Cleaners Mem. Fac. Technol., Tokyo Metrop. Univ., (No.
on Properties of Heat-Treated Medium- 15), 1965, P 1163-1176
Carbon and Low-AlloySteels, Heat Treatment 74. RE Kern, Selecting Steels for Heat-Treated
'87 (London), Instituteof Metals, 1987 Parts, Met. Prog., Vol 94 (No.5), Nov 1968, P
59. WR McFarland and 1.T. Cronn, Spheroid- 60-73
isation of Type IT Manganese Sulfide by Heat 75. L.O. Uhrus, Clean Steels, Special report 77,
Treatment, Metall. Trans. A, Vol 12 (No.5), The Iron and Steel Institute, 1963, p 104-109
May 1981,p 915-917 76. E Vodopivec and L. Kosec, The Fatigue
60. R Kiessling, Nonmetallic Inclusions in Steel Strength and the Influence of Nonmetallic In-
Ill, Metals Society, 1978 clusions on the Fatigue Fracture of
61. S.I. Gubenko, Phase Transformations in Non- Case-Hardening Steels(BISI5906), Hart. -Tech.
metallic Inclusions during Heat Treatment of Mitt., Vol 22, July 1967, P 166-173
Steels,Izv. V.U.Z ChernayaMetall., (No. 12), 77. S. Gunnarson,StructureAnomaliesin the Sur-
1986, P 67-71 face Zone of Gas-Carburized, Case-Hardened
62. S.I. Gubenko, Dissolution of Nonmetallic In- Steel, Met. Treat. Drop Forging, June 1963, p
clusions during High Temperature Heating, 219-229
Izv. Akad. Nauk SSSR,Met., (No.2), Mar/April 78. R Winter, G. Knauer, and 1.1. Gamel, White
1983, p 103-107 Etching Areas in Case Hardened Gears, Gear
63. W.A.Spitzig and RJ. Sober, Influence of Sul- Technol., Sept/Oct 1989, p 18-44
fide Inclusionsand PearliteContent on the Me- 79. WE. Littman and RL. Widner,Propagationof
chanical Properties of Hot-Rolled Carbon Contact Fatigue from Surface and Sub- Sur-
Steel, Metall. Trans. A, Vol 12, Feb 1981, p face Origins, Trans. ASME, Sept 1966, p
281-291 624--636
64. G.E. Dieter, Mechanical Metallurgy, Metal- 80. K Suginoet aI.,StructuralAlterationsof Bear-
lurgy and Metallurgical Engineering Series, ing Steels under Rolling Contact, Trans. lSI],
McGraW-Hill, 1961, p 270-271 Vol 10, 1970, P 98-111
65. G. Kiessling, NonmetallicInclusions in Steel V, 81. G.R Speich and W.A. Spitzig, Effect of Vol-
Metals Society, 1989 ume Fraction and Shape of Sulfide Inclusions
66. D. Brooksbankand K W Andrews,Production on Through-Thickness Ductility and Impact
and Application of Clean Steels, J Iron Steel Energy of High-Strength 4340 Plate Steel,
Inst., Vol210, 1972,p 246-255 Metall. Trans. A, Vol 13, Dec 1982, p
67. KB. Grove and J.A. Charles, Further Aspects 2239-2259
of InclusionDeformation, Met. Technol., Vol 1 82.1.S. McBride, Overheating-Reappraisal of a
(No.9), Sept 1974,p425-431 Reccurring Problem, Met. Mater., Vol 8 (No.
68. RN. Cummings, EB. Stulen, and We. Schulte, 5), May 1974, p 269-270
Tentative Fatigue Strength Reduction Facors for 83. RW Gardiner, Effect of Overheating on the
Silicate-Type Inclusions in High-Strength Steels, Fatigue Strength and Other Mechanical
Proc. ASTM, \b158, 1958, P 505-514 Properties of Carburized and Uncarburized
69. T. Ohta and K Okamoto, Effect of Nonmetal- VAR S 82 Steel, Met. Technol., Dec 1977,p
lic Inclusions on Fatigue Life of Bearing 536-547
Steels, Tetsu-to-Hagane (J Iron Steel Inst.), 84. P. Hart, Effectsof Steel Inclusions and Residual
Vol52, 11, Lecture 104, 230, 1966 Elements on Weldability, Met. Constr., Oct 1986
70. T. Tsunoda, I. Uchiyama,and T.Araki,Fatigue 85.1. Hewitt, Hydrogen in Steel, B.I.S.R.A.ffhe
(of Steels) in Relation to the Non-Metallic In- Iron and Steel Institute, 1961
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Carburizing: Microstructures and Properties Copyright © 1999 ASM International®
Geoffrey Parrish, p 135-170 All rights reserved.
DOI: 10.1361/cmap1999p135

Chapter 6

Core Properties and Case Depth

When a case-hardened machine part is sub- neath the case can be significant. Under these
jected to normal loading, its failure, should fail- conditions, the core ceases to be 'just stuffing."
ure occur, usually results from contact damage or
damage due to bending stresses, possibly acting
on an engineering or metallurgical stress raiser.
Contact damage often initiates at the surface as a Core Factors
consequence of frictional effects (predominantly
sliding and wear processes), and, therefore, the The core properties of a carburized part are
surface condition and the metallurgical structure dictated by the chemical composition of the steel
of the surface are very much involved. Contact and the rate at which the part cools during
damage can also develop at subsurface locations quenching. The alloy content is mainly responsi-
due to shear stresses, which are generated below ble for the depth to which a steel will harden
the surface when one surface rolls over another. (hardenability), whereas the carbon content
In this situation, the metallurgy of the material largely determines the hardness.
beneath the surface (where the maximum shear
stresses develop) needs to be considered. When Core Hardenability
bending conditions prevail, as at a gear tooth fil-
let, the stresses developed are greatest at the sur- Case-hardening steels are usually lean alloy
face and decrease steeply beneath the surface; with total alloying-element contents ranging
again the surface and near-surface metallurgy are from about 1 to 6.5% and carbon contents be-
important. Therefore, if the case depth of a tween about 0.1 and 0.25%. When the total alloy
case-hardened part is deep enough, the material range is 4 to 6.5%, the main alloying element is
of the case is strong enough, and the load-bear- usually nickel. The available case-hardening
ing areas are large enough, the notion that the steels can produce as-quenched core hardnesses
core is 'just stuffing" is more or less valid. Un- between 20 and 45 HRC (depending upon size),
fortunately, it is not that simple. In general, eco- and are suitable for case-hardened parts from a
nomic and ecological considerations dictate the few millimeters to almost a meter in section.
use of the leanest (or cheapest) steel for the job, Case-hardening steels for special applications,
as well as the shortest time in the furnace needed for example, high-temperature service applica-
to produce just enough carburized case to cope tions, may contain at least 10% alloying ele-
with the applied stresses in a minimum ments. Such steels are generally tool steels
weight-to-power ratio situation. The aim is to not adapted for carburizing and require special heat
over design (to do so increases the cost); there- treatments to develop desired properties. There-
fore, there is a risk of under designing with re- fore, they are not directly discussed in this work.
spect to material selection and case depth speci- When selecting a steel for an application, two
fication. This is designing nearer to the limit, and important requirements must be satisfied. First,
in such circumstances, the contact and/or bend- the steel must have adequate case hardenability
ing stresses experienced by the core material be- for the job; that is, the carbon-rich layer induced
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
136 / Carburizing: Microstructures and Properties

by carburizing will suitably harden when section size is large and the steel has low
quenched. A number of steels might satisfy this hardenability. In such an instance, the case might
requirement, but each will likely have a different not satisfactorily harden. One reason ferrite
core hardness range. Therefore, the second re- could be present in an as-quenched core material
quirement to be satisfied is core hardness (or is that it was austenitized (prior to quenching) at
strength) of the critical core areas of the compo- a temperature below the AC3 temperature, leav-
nent. This primarily relates to the hardening ing some ferrite undissolved. Another reason,
depth (the core-hardenability). Figure 6.1 pro- perhaps the most common, is that a steel is em-
vides a general view of core hardenability in ployed that has less hardenability than required
terms of how average tensile strength varies with for the section. This is illustrated in Fig. 6.3
section size and alloy content. where the upper continuous-cooling transforma-
Microstructures observed within the cores of tion (CCT) diagram represents a lean-alloy
case-hardened parts are: ferrite (undissolved or case-hardening steel, and the lower diagram
precipitated), bainite (upper and lower), and represents a medium-alloy case-hardening steel.
martensite (low-carbon) (see Fig. 6.2). Pearlite is By superimposing near-surface cooling curves
not found in a quenched core material unless the (e.g., for -12 mm and -50 mm diameter bars),

Diameter, in.

1391 Tolerance ±7 tsi Tolerance ± 10 tsi


835 M 15
8!. 1082
822 M 17
:::iii 659 M 15
c 820M 17
U; 927 832 M 13

773 50

4 6 4 6 10
Equivalent diameter, in. Equivalent diameter, in.
I I I ~I--IL...-*"".......~b-.......~;;-
I .......~~~~I
o 50 100 150 200 250 0 50 100 150 200 250
Equivalent diameter, mm Equivalent diameter, mm

Nominal composition, \I>

Steel C Mn Ni Cr Mo
835M 15 0.15 0.40 4.1 1.1 0.20
659MI5 0.15 0.40 4.1 1.1
832M 13 0.13 0.40 3.2 0.90 0.20
655M13 0.13 0.50 3.2 0.90
655M 17 0.17 0.50 1.8 0.25
822M 17 0.17 0.60 2.0 1.5 0.20
820M 17 0.17 0.80 1.8 1.0 0.15
815MI7 0.17 0.80 1.4 1.0 0.15
637M17 0.17 0.80 1.0 0.80 0.05
635MI5 0.15 0.80 0.80 0.60 0.05

Fig. 6.1 Strength versus section diameter for a number of U.K. carburizing steels. Use lower scale to estimate bar cen-
ter strength; use upper scale to estimate bar surface strength. Source: Ref 1
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Core Properties and Case Depth / 137

8oor-~-........::---~--------, it is seen that for such conditions, the formation

700 of ferrite is unavoidable in the leaner grade
o 600 but avoidable in the more alloyed grade. Conse-
i.a SOO I:=-==-=,,=~
quently, if ferrite is not desired in the core of a
as part, an adequately aIIoyed steel must be selected
lii 300 with quenching from the fully austenitic condi-
~ 200 h¥iiwJ-4-.../ (e)
tion. Increasing the cooling rate might suppress
o ferrite formation up to a point. The selection of a
-looL..-_--l.._ _...J.... _ _l--_--l.._ _-l-J borderline grade of steel might, on a
1 10 105 batch-to-batch basis, lead to a fair amount of
lime, S core structure variability. This is because
steelmakers cannot work to precise chemical
compositions; they must have reasonable work-
ing tolerances. In the unlikely situations where
each alloying element in a steel is either at the
bottom or at the top of its specification range, the
transformation characteristics can be extremely
variable, as Fig. 6.4(a) and (b) suggest. Even
variations of carbon content alone can make a
difference in transformation behavior (Fig.
6.4c ). This indicates for this steel that if the
carbon content is low at 0.08%, the time to the
ferrite nose is less than 8 s, which means, for

~ ~ ........ .......F
M /"8
\. ~ Jdiam
~"\ 12 mm "- ~Mn bar - 5i
200 diam - Ni 0.50%
100 bar_ I------ Cr 0.50"10
Mo 0.20%
1 10 1()2
(b) Time,s
800 C 0.12%
....... 51 0.20"10

~ -, Mn 0.50%
~ 600
" ............... F Ni 3.20%
~ 500
M '" \.1"8 <, -
Cr 0.85%
Mo 0.12%
\-;2m~mm- "
dia~ _diam
bar bar
1 10
(c) lime, S
Fig. 6.2 Microstructures obtained by cooling a 0.16%C- Fig. 6.3 A comparison of the continuous-cooling trans-
3%Ni-Cr steel from 920°C. (a) Fast cool (920-200 °C in 30 formation diagrams for (a) BS 970 805M20 (SAE 8620)
s) giving low-carbon martensitic structure of 1590 MPa (compositio n: 0.17 C, 0.30 st. 0.80 Mn, 0.50 Ni, 0.50 o.
UTS. 800x. (b) Intermediate cool (920-250 °C in 200 s) giv- 0.20 Mo) and (b) BS 970 832M13 (composition: 0 .12 C,
ing ba initic structure of 1360 MPa UTS. 800x. (e) Slow cool 0.20 Si, 0.50 Mn, 3.20 Ni, 0.85 Cr, 0.12 Mol. F, ferrite; B,
(920-250 OC in 10" s) giving a ferrite/pearlite structure of bainite, M, martensite. Surface cooling is shown for a 12
740 MPa UTS. 800><. Source : Ref 1 mm diam bar and a 50 mm diam bar.
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
138 / Carburizing: Microstructures and Properties

example, that an oil-quenched 12.5 mm (Yz in) as-quenched core strength of less than about
diameter bar will contain some ferrite at its cen- 1000 MPa (145 ksi). With this in mind, one can
ter. When the carbon content is raised to 0.18% carburize, slow cool, and subcritical anneal to fa-
and the other elements remain the same, the time cilitate the removal of areas of the case where,
to the ferrite nose is nearer to 200 s. This allows after quenching, further machining operations
a 125 mm (5 in.) diameter bar to be oil quenched can be performed.
without any ferrite production. Fortunately, steel
manufacturers can generally maintain the
amount of each element in a steel grade within
narrow limits. The purchase of controlled- 10001==============1
hardenability steels also makes for a greater de-
gree of consistency from batch to batch. In terms o 7001-

of end-quench hardenability, Fig. 6.5(a) and (b)
~ 500
show the harden ability curves that correspond to
the CCT curves of Fig. 6.3; the extremes of the E300
8. 400'----
{Eo 200
bands of Fig. 6.5(b) correspond to the CCT 100
curves of Fig. 6.4(a) and (b). o......---:':---~;;---~;;--~.-----:7.
1 10 10 103
There are areas within the core of a carbu- Time, S
rized and hardened component that experience (8)
little, if any, stress during service; for these ar-
eas, the microstructure and strength are of little 900
concern. It is those areas just beneath the case u 800 ......
0_ 700
where high stresses develop during service for
which core structure and strength are important.
soo '\ <:..
Therefore, when using CCT diagrams to assess 8. 400 -
the suitability of a steel for a given component,
E 300
{Eo 200 \
the near-surface cooling rate is the most mean-
1 10
102 103
Ferrite cores, although not uncommon, are of- Time,s
ten regarded as unacceptable for critically loaded
components. Bainite-type cores are perhaps 1000r---------------,
more common and more desirable than ferrite. 900
The martensitic, or predominantly martensitic, 800
core structures tend to be found in those compo-
~ 700
nents with small sections, and these are also de-
f!! 600
sirable if the core carbon and the core hardness :::J

are not too high. ~ 500

In selecting a steel for a particular component, ~ 400~~~~1I=w.~-....._ _
care must be taken to choose one without too ~ 300
much carbon and hardenability for the shape and 200.~~E.",-­
the section involved, otherwise growth, distor- 100
tion, and internal cracking (as under the tops of O~-- _ _'_ _'~ ____'____'
gear teeth) might become a problem. Also, one 1 10 103
needs to decide whether to choose a steel of rea- (e)
sonable hardenability and harden it with a mild
quench, or to adopt a lower hardenability steel Fig. 6.4 Continuous-cooling trans~ormation diagra~s
and hard quench it. The choice between a low for selected 3%Ni-Cr case-hardening steels. Specifi-
hardenability steel with a high carbon content, cation En 36 is now replaced by 655M13 and 831M13.
(a) Ni, Cr, and Mo contents all atthe bottom of the specifi-
and a higher hardenability steel with a low car- cation range (En 36). Composition: 0.12 C, 0.20 Si, 0.40
bon content often depends on the duty of the fin- Mn, 3.00 Ni, 0.60 Cr, 0.00 Mo, 0.00 V. (b) Ni, Cr, and Mo
ished item. However, a low hardenability steel contents all at the top of the specification range. Compo-
with a high carbon content is more prone to size sition: 0.12 C, 0.20 Si, 0.60 Mn, 3.50 Nil 1.1 0 Cr, 0.25
Mo, 0.00 V. (c) Effectof carbon content with alloying ele-
variations. One good reason for selecting an al- ments at constant levels. Composition: 0.06-0.18 C, 0.20
loy case-hardening steel with a low nominal car- Si, 0.54 Mn, 3.18 Ni, 0.91 Cr, 0.04 MOl 0.00 V.Source:
bon content (-0.13%) is that it should develop an Ref 1
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Diameters of rounds with Quench
sameas-qllenched hardness (HRC), in. Location in round

2 4 Surface Mild
1 2 3 4 :Y4radius from center water
0.5 1 1.5 2 2.5 3 3.5 4 center quench

1 2 ~ 4 Surface Mild
0.5 1 1.5 2 2.5 3 3.5 4 :Y4radius fromcenter oil
0.5 1 1.5 2 2.5 3 3.5 Center quench




Heat-treating temperatures recommended by SAE
Normalize(for forgedor rolled specimensonly): 925°C (1700 oF)
50 Austenitlze: 925°C (1700 oF)

,\ -,
\ -,

o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Distance from quenched end,1/16in.

Diameters of roundswith Location in round Quench

sameas·quenched hardness (HRC), in.
2 4 Surface Mild
1 2 3 4 3/4 radiusfromcenter water
0.5 1 1.5 Center quench
2 2.5 3 3.5 4

1 2 3 4 Surface Mild
0.5 1 1.5 2 2.5 3 3.5 4 3/4 radiusfromcenter oil
3.5 center quench
0.5 1 1.5 2 2.5 3

55 Heat-treating temperatures recommended by SAE
Normalize(for forgedor rolled specimensonly):925 OC (1700 oF)
50 Austenitize: 845°C (1550 oF)
:I: 45

~ -r-- ' -


- <,
r-- "'--

. . . r-- r--

o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Distance fromquenched end,1/16in.
Fig. 6.5 Hardenabilityrangesfortwo case-hardening grades of steel. Source: Ref2
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
140 / Carburizing: Microstructures and Properties

Core Microstructure and Hardness trol is tight. In that way, the attainment of a mini-
Core hardness is sometimes specified on engi- mum core hardness of the finished part is virtu-
ally ensured.
neering drawings for case-hardened parts, im-
plying that a heat-treated part is to be sacrificed From a designer's point of view, a hardness test
to permit the necessary sectioning and hardness value is only a means of conveying what is re-
testing. In such instances, the hardness tests are quired or what has been achieved; strength data
carried out as a standard procedure in which test are far more meaningful to the designer. Fortu-
impressions are made at specified locations. For nately, the relationship between hardness .test val-
gear teeth, these might be at the center of the ues and equivalent tensile strengths is quite good
(Fig. 6.6), and it is possible to approximate other
tooth and at the center of the tooth on the root
circle diameter. Such tests relate the minimum properties from the tensile strength. .
tooth hardness, which has some value even The connections between strength properties
though, in the locations specified, the service- and microstructure tend to be somewhat blurred.
This is because core microstructures are often a
applied loads are negligible. Perhaps a more
meaningful test site would be at, for example, mixture of different phases; cores with
2 x the total case depth at the midflank and tooth hardnesses of <25 HRC will have high ferrite
fillet positions. Hardness tests are quick and easy contents, whereas >40 HRC cores would indi-
to perform and, as a rule, are reliable. cate a predominantly martensitic microstructure.
When it is unacceptable to determine the core Hardness values of mixed microstructures are
hardness of actual parts, the best alternative is to between 25 and 40 HRC. The effect of structure
variability on the hardness of an alloyed
have a simulation test piece that contains appro-
case-hardening steel is illustrated in Fig. 6.7;
priate geometric features and cools during
note that other steels with different core carbon
quenching at a rate similar to that of the actual
and alloy contents will shift the lines of the dia-
part. The third option is to ensure that the mate-
rial selection is related to the size of the compo- gram to the left or to the right along ~e hardness
scale, depending on the actual cherrucal compo-
nent (Fig. 6.1 or equivalent), that the material
supplied is to specification, and that process con-

Core Tensile Properties


2400 L Core Tensile Strength. The influence of core

2200 L carbon on the center tensile strength of 3%Ni-Cr

steel parts with different sections is shown in Fig.
6.8. This figure suggests that there can be a 310 to
460 MPa wide band on the ultimate tensile
~ 1800 / strength (UTS) value due simply to a variation of
::i: carbon content within a range of 0.08% C.
Normally, for case-hardening steels, the tolerance
~ 1400
L band for carbon content is 0.05 or 0.06% C, and
Q) steelmakers can readily achieve this, as the inset il-
~ 1200
c: lustration in Fig. 6.8 shows. Nevertheless, some
1000 / core strength variability can be anticipated due to
/ carbon and composition variations. For many pro-
duction parts, the center core strength is not too im-
V portant, whereas that part of the core irnm~a~ly
/ beneath the stressed case is important. An indica-
tion of how hardness and, hence, strength vary
0 within a section is illustratedin Fig. 6.9.
20 0 100 200 300 400 500 600 700
Core Yield Strength. From a designer's point.of
Hardness, HV
view, the core yield strength often has more Sig-
nificance than the tensile strength does because,
20 253035 40 45 50 55 60
Hardness, HRC in most designs, core yielding is not permitted.
Fig. 6.6 Strength and hardness conversion. 1 psi = There are different ways of conveying yield data,
0.00689476 MPa. Source: ASM Metals Reference Book, for example, limit of proportionality (LoP),
1981 proof stress (offset yield), or yield strength. In
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
oJ!. 1.0 0.16 -+-+-+---1-l-+--+-+-+--+~-+
E 1.50.15 0.26
~ Fr--+-++-+:':;:-+-If-t-+-+~---l
NotII: Martensite hardness is approximately 45 HACfor a 4%alloysteel 8 2.0 0.140.15 0.25
wi1h 0.16%C. Forleaneralloysteels, onemoves vertically up thealloy
1;' 3.0 0.13 0.15 0.24
contentscale for equivalent hardness. Forexample, a 1.0%alloycontenl
::e 4.0 0.13
wouldrequire 0.19%C for a martenstte hardness of 45 HRC. 0.23
100 r----------~.-----------''-''':::.L....J-...l...-.,,~L......J----I.--L--'----'--'-__'__i


~ 60
Changes in
__-I~_ carbon content
40 increase or
20 for 100% martensite.
200 250 350 400 450 500
Hardness, HV
22 25 35
30 40 45 48 50
Hardness, -HRC
Fig. 6.7 Approximate relationship between core microstructure and hardness of a Ni-Cr-Mo carburizing
steel (-4% alloy content) with -0.16% C. The alloy content/carbon content extension (top right) permits the
phase % plots to be "moved" in relation to the fixed hardness scale to approximate core strength for other
steels in slide-rule fashion. Below -250 HV represents slow-cooled (normalized) and annealed materials,
and therefore, bainite could read as bainite, pearlite, or spheroidized carbides. Above -250 HV refers to
quenched materials. For the 180°C tempered condition, there will be zero change at 360 HV and below, but
there will be a 20 point HV loss at 100% martensite.

Tensilestrength, ksi

135 180

60 59


Q; 40
E 75
z 30
.5 E
10 i
«l 0 50 E
0 0.11 0.12 0.13 0.14 0.150.16 0.17 0
Carbon, %


0.08 0.10 0.12 0.14

1200 1400

Tensile strength, MPa

Fig. 6.8 Effectof section size and carbon content on the strength of oil-quenched 3%Ni-Cr carburizing
steels 832M13 and 655M13. Source: Ref 1
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
142 / Carburizing: Microstructures and Properties

~ < , .[
/ 300

200 815 M17 steel

Fig. 6.9 Hardness distribution within a toothed section. Source: Ref 1

Fig. 6.10 for UK steel 822M17, the LoP is at as in martensite-bainite structures. The smaller the
about 50% of the UTS, the 0.1% proof stress is packet size or lath width, the higher the yield
at -60%, the 0.2% proof stress is at -70%, and strength and toughness are (Ref 4).
the AGMA yield number is about 80% of the UTS. Another way of looking at yield strength data
Yield strength is influenced by microstructural is with the yield ratio (YR), that is, the UTS di-
constituents, grain size, or packet size or lath width, vided by one of the yield strength indicators. In

1800 ,.-------,-----,.----,------,.-------,-----,.------,

1600 49

Limit of
1400 45


1200 39 .,'"c:<Ii
c: "E

1000 32

Steel: U.K. 822 M 17

800 23

600 '--_ _--'- --'- ...L- L -_ _--'- --'-_ _- - - '

400 600 800 1000 1200 1400 1600 1800

Proof stress, MPa
Fig. 6.10 Relationship between ultimate tensile strength and proof stress for a Ni-Cr-Mo carburizing steel. Derived
from Ref 3
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Core Properties and Case Depth / 143

Fig. 6.11, the yield ratio UTS/0.2% proof Core Ductility. The ductility indicators are the
stress. This illustration represents bars of 1X in. percent reduction of area and the percent elonga-
diameter with carbon contents up to 0.18%, and tion as derived from the tensile test; the higher
shows that for core strengths typical of the values of the two indicators, the more ductile
martensitic and/or bainitic microstructures, the the metal is. Ductility relates to the ability of a
YR is fairly low (around 1.3 to 1.5); that is, the material to be plastically deformed without frac-
yield strength is fairly high. With smaller section ture; it is important to those who work metals,
sizes and carbon contents above 0.18%, yield ra- for example, wire manufacturers or car-body
tios down to 1.15 are possible. Core materials press shop operators. For case-hardened parts,
containing ferrite tend to have low yield the significance of core ductility is not at all clear
strengths and high yield ratios (1.5 to more than because the applied stresses must exceed the en-
2.0); how high depends on the amount of ferrite gineering yield strength of the core before ductil-
present. (The significance of a fairly high-yield ity becomes a consideration. Nevertheless, ac-
strength core material will become more appar- ceptance testing for case-hardening steels
ent when fatigue resistance is considered.) Inter- requires that the percent reduction of area, the
estingly, regarding straining in tension, if the percent elongation, and the impact resistance are
yield ratio is more than 1.4, work hardening will determined, if only to assure the manufacturer
occur during cyclic straining. If, on the other that the steel is of acceptable quality. According
hand, the ratio is less than 1.2, work softening to the trend, as the strength and hardenability of
will occur during cyclic straining. Between 1.2 the steel increase, the ductility decreases (Fig.
and 1.4, the material is probably fairly stable 6.12).
The yielding referred to so far is macroyielding Core Toughness
as determined during a typical tensile test. How- Toughness is the ability of a metal to absorb
ever, before that stage is reached, microplastic energy. This is generally important with respect
yielding has already commenced at a much lower to case-hardened parts because tough case-
stress. The onset of microplastic yielding, the true hardened parts are more able to survive occasional
elastic limit, is when dislocation migration first
occurs, and can be measured by determining the
change of AC resistance while straining in uniax- 245
ial tension (Ref 6). Tests on samples of a Proof stress
through-hardened and low-temperature tempered 'w
case-hardening steel have indicated that the true vi
elastic limit is only a few percent below the bend- !!!
u; 190
ing fatigue limit.
Strain ",0.5%C
softening Strain 135
I hardening I
1600 225 40
&. 1500 f---f-- 'w
~o Reduction of area
:2 1400 f---t- 200 ~ ;fl. 30 0

~ i5> o .~
1300 f---f--
:::cI n 20
~ 39
1200 f---f-- u; 8-
<, ~ _ 0•
~ 1100 1--1----' 150

'1jj 33
1Il 10
~ ,-'"
Q) v IA-.", : _
~ 1000 1 - - - + - c:
Elongation I<Y •
"E "1f' I-
900 f---t--'
125 -@ 26 :::c
-@ 800 I-----t--+---' E

700I-----t--+--I---1---=I 100
600 L..----'_---'_--'-_--'-_-'
:5 .
;§ 20

- 0 .n
1.0 1.2 1.4 1.6 1.8 2.0 ,I
Ratio of ultimate tensile stress to 0.2% proof stress
165 195 225 255 285 315
Fig. 6.11 Relationship between the ultimate tensile
strength and the 0.2% proof stress (offsetyield) of carbu- Tensile strength. ksi
rizing steels (0.08-D.18% C), Note that with carbon con- Fig. 6.12 Mechanical properties of lightly tempered
tents of over 0.18%, the ratio can be as low as 1.15 for plain-carbon martensites to illustrate how ductility falls
strengths over about 200 ksi (44 HRCl. Data from Ref3 as strength rises. Source: Ref 7
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
144 / Carburizing: Microstructures and Properties

overloads and cyclic impact loading than are test has largely been driven by the national
parts of relatively low toughness. standards, and its value, if any, is regarding mate-
Tests for toughness include notched (severe rial acceptance. Even used for that purpose, a
to gentle notches) and unnotched impact bend notched impact value of 20 ft . lb (27 1) is con-
tests, notched and unnotched slow bend tests, sidered ample for most applications, though the
and precracked fracture tests. Such tests indi- appropriate material standard might call for
cate the contributions of alloying elements, more than 27J (Ref 9). Another limitation of the
microstructure, and hardness to fracture resis- standard test is that the results do not relate to fa-
tance. The more meaningful information is ob- tigue performance, nor can they be used in de-
tained not from the standard impact tests sign formulas. The more recently adopted instru-
(Charpy and Izod) but from those tests in which mented impact and bend tests used for research
the stress raiser in the test piece is more similar projects, while still having some limitations, em-
to those found in actual machine components. ploy test pieces that contain more realistic stress
Chesters examined the limitations of the stan- concentrators; therefore, these tests more use-
dard impact test in an early work (Ref 8). The fully contribute to the understanding of the re-
results of this research for a number of steels spective influences of steel composition, core
are summarized in Fig. 6.13. The standard test microstructure, and hardness (strength).
on its own did not supply much information on Fett showed that steels containing little or no
the effect of material composition on impact, nickel are notch sensitive at core hardnesses of
whereas the simulated gear tooth did. Further, >40 HRC in slow bending, and sensitive at >30
even with Izod impact test values below 10 HRC under impact loading (Ref 10). Nickel
ft . lb, the case-hardening grade of steel (En 36) steels (>1.8% Ni) are not so sensitive at that
still offered a good resistance to impact when value, which means that with the nickel grades of
the stress concentrator was similar to that for a case-hardening steels one can develop a
small gear tooth. The use of the notched impact high-yield strength core that is adequately tough.

lzod, J
80 r---....,..;::.-,---;;----r...::;::::-,----:;:i----r--T=---,----T.,-----,

60 f-~~r---+---+ Medium carbon -+-----=::-+---+--=001.-.=:.----t---'------i 80

<= 50
E 60 '0
a 0 ~
en 40 0
0 Q.
Q. .E
.5 OJ
jg 30 40 '0
a Positionof blow a
20 o Ni
• Medium carbon 20
.. En 19
A En 36
o En 29
• En 30 l'lOmm2 ,I
°O!----:l=--~:I=---::!~---!:---==-----:!::---:::.::----=--~--7,loo o
30 40 50
lzod, It ·Ibf
Fig. 6.13 Comparison between standard impact test results and results from a test piece designed to simulate a gear
tooth. En 19, 705M40 (nominally, OAOC, 1.0Cr, 0.3 Mo); En36, 832M13 (nominally, 0.13 C, 3.2 Ni, 1.0Cr,0.15 Mo); En
30, 835M30 (nominally, 0.30 C, 4.0 Ni, 1.2o.
0.3 Mo); En29, 722 M24 (nominally 0.24 C, 3.0 o.
0.5 Mol. Source: Ref8
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
Core Properties and Case Depth / 145

The higher nickel grade (>3%) generally has su- from a common Cr-Mn-TI steel base stock with
perior toughness properties compared to the me- carbon contents between 0.07 and 0.45%. The
dium nickel grade (-1.8%), though the latter is residual stress distribution curves obtained are
still good. However, in practice, a steel is se- shown in Fig. 6.14(a), which shows that as the
lected for its case and core hardenabilities, so the carbon content of the core material increases
tendency is to use the more alloyed grade for the (and, hence, the hardness increases), the value of
larger components. The effect of steel composi- the surface compressive stress decreases at a rate
tion was determined by Cameron et al. who rated of 11 MPa per 0.01% C (Fig. 6.14b). The peak
the as-carburized impact strength of Mn-Cr compression (here, -75% of the case depth) fol-
steels as less than that of Cr-Mo steels, which, in lows the same trend with respect to carbon, apart
tum, were somewhat inferior to a Ni-Cr-Mo (pS from the lowest carbon level (Fig. 6.14c). The
55) steel (Ref 11). For the Mn-Cr and the Cr-Mo same researcher reported that similar tests with a
steels, increasing core carbon from about 0.17 to
3.5Ni-1.5Cr steel produced the same trend, ex-
0.3% decreased fracture strength, but a similar
cept the values of maximum compressive stress
variance of core carbon had no effect on the frac-
were very much higher, that is, 0.18, 0.28, and
ture strength of the PS 55 steel. The beneficial
affect of molybdenum with nickel on fracture 0.38% core carbon samples gave values of 830,
strength was demonstrated by Smith and 860, and 730 MPa, respectively. The trend found
Diesburg (Ref 12); they showed that increases of by Sagaradze was more or less confirmed by
both nickel and molybdenum raise the toughness Kern whose curve for the SAE 8600 series of
of case-hardened test pieces. steels is also shown in Fig. 6.14(c) (Ref 14).
Effect of Core Strength on Bending Fatigue Re-
sistance. Increasing the core carbon increases
Effects of Core Properties the core strength for a given quench or micro-
structure, and increasing the quench severity in-
Effect of Core Material on Residual Stresses. The
creases the hardness (up to a point) for a given
residual stress distribution within a case- hard-
carbon content. Increasing the core strength can
ened layer is related to the difference in volume
expansion between the high-carbon martensite of reduce the amount of compressive residual stress
the case and the low-carbon martensite, bainite, within the case. A reduction of surface compres-
or ferrite of the core (Ref 1). The greater that dif- sion can then lead to a reduction of fatigue resis-
ference is, the greater the likelihood is of produc- tance, shown in Fig. 6.15. From this, it can be
ing high magnitudes of residual stress, provided concluded that there is an upper limit of desir-
no yielding occurs and the sequence of transfor- able core strength for case-hardened parts.
mation is correct. This implies that a part with a Low-carbon cores, while they encourage the de-
low-carbon core should develop a more favor- velopment of high surface compressive-residual
able residual stress distribution than a part with a stresses, might either deform under load or lo-
high-carbon core. Sagaradze (Ref 13) carried out cally yield enough to modify the residual stresses
residual stress determinations on carburized and and increase the possibility of subcase fatigue-
hardened plate samples that had been prepared crack initiation. In other words, for critically

196 196 20 0 0


~ 0
ui 0

ui -20
l Cr-Mn-Ti

en ~

~ ~
en -294 en -30
! 0

-294 -30 -196 -20 -392 -40 l...-~---L_I.--~.......J.---J

o 0.2 0.4 0.6 0.8 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0.5 0.6
(a) Depth,mm (b) carbon (core), % (c) Carbon (core), %

Fig. 6.14 Dependence of residual stress in carburized and hardened cases on core carbon level. (a) Residual stress dis-
tribution in samples of Cr-Mn-Tisteel of varying core carbon contents; case depth, 1.2 mm; quenched from 810 0c. (b) Re-
lationship between surface residual stress and core carbon. (c) Relationship between peak compressive stress and core
carbon. Source: Ref 1 3, 14
‫ﻣﺮﺟﻊ ﺩﺍﻧﺸﺠﻮﻳﺎﻥ ﻭ ﻣﻬﻨﺪﺳﻴﻦ ﻣﻮﺍﺩ‬
146/ Carburizing: Microstructures and Properties

Core carbon, %
0.10 0.20 0.27 0.36 0.45
Core strength, HRC
15 31 43 49 53

774 N
0.. 872
~ -§,
~ 774
725 ~ c
tc Notched
Vi 676
~ 1;; 70 samples
~ OJ
:::l .. '
Ol :::l
$ 588 Ol
60 •••••• ::: •••
40 Kh ........ ...... ........ .......
~ 676 Ol
LL ~
C Ol ....... ....... .. .. .....
'6 490 c 50 ..
c '6
OJ c "20 KhNM
Compressive Tensile CD OJ Tensile
627 65 392 40 '---.1-:,-----'---:'-:---.1-:,-----'
-20 0 +20 o -20 -40 --60 -80
Surface residual stress, kg/mm2 Surface residual compressive stress, kg/mm2
-196 0 +196 o -196 -392 -588 -774
Surface residual stress, MPa Surface residual compressive stress, MPa
00 ~
Fig. 6.15 Relationship between fatigue limit and surface residual stress for the Cr-Mn-Ti steel referred to in Fig. 6.14.
Generally (a) A reduction in surface compressive stresses leads to (b) A reduction in bending fatigue resistance. Source: Ref

loaded parts, there is a lower limit of desirable further. Four of the sources (Ref 13, 16, 17, 20)
core strength. indicate an optimum core strength of -1150 to
There have been several attempts to relate the 1200 MPa (37-39 HRC), whereas two other
experimentally derived fatigue limit to the ultimate sources (Ref 18, 19) report 1250 to 1550 MPa
strength or hardness of the core material in order to (40-48 HRC). In terms of microstructures, one
arrive at the optimum value of core strength. Fig- might expect a predominantly bainitic structure
ure 6.16 shows that the fatigue strength rises with with less than 50% martensite and no ferrite for
core strength up to a peak value, beyond which the the 37 to 39 HRC range; for the 40 to 48 HRC
fatigue strength falls as the core strength increases range, the structures will contain 50 to 100%

Approximate UTS, MPa

1000 1200 15001800 Core strength, MPa Core strength, MPa
~o 1.6m~872 '"
620930124015451855 310 620 930 124015451855
60 ..
N 90