Available online at www.sciencedirect.

com
International
Journalof
International Journal of Fatigue 30 (2008) 1441–1447
Fatigue
www.elsevier.com/locate/ijfatigue

Fatigue strength and fracture mechanism of different post-fused

thermal spray-coated steels with a Co-based self-fluxing alloy coating
a,*
Jeongseok Oh , Jun Komotori b, Jungil Song c

a
School of Mechatronics Engineering, Changwon National University, Changwon, Kyungnam 641-773, Republic of Korea
b
Department of Mechanical Engineering, Keio University, 3-14-1 Hiyoshi, Kohuku-ku, Yokohama-city, Kanagawa 223-8522, Japan
c
Department of Mechanical Engineering, Changwon National University, Changwon, Kyungnam 641-773, Republic of Korea

Received 18 January 2007; received in revised form 15 October 2007; accepted 15 October 2007
Available online 22 October 2007

Abstract

The characteristics and adhesive strength of thermal-sprayed coatings at the interface between the coating and substrate are affected
by the post-fusing treatment process. We investigated the effect of the adhesive strength on the fatigue strength and fracture mechanism
of post-fused specimens. Rotating bending fatigue tests were conducted on specimens with a Co-based self-fluxing alloy coating on a
medium carbon steel substrate. The post-fusing treatment was performed using a vacuum furnace, an electric furnace, or an induction
heating system. A diffusion layer formed at the interface of the specimens treated in either furnace at 1100 C for 4 h. These conditions
produced a strong adhesive strength; the fatigue strength of these specimens also increased remarkably compared to the substrate-only
specimens. However, the specimens treated with an induction heating system at 1100 C for 120 s had a much lower adhesive strength
since a diffusion layer was not formed, leading to delamination of the entire coating from the substrate during the first stage of the
fracture process. The fatigue strength of these specimens was almost equal to that of the substrate-only specimens. Thus, thermal-sprayed
coatings treated in vacuum and electric furnaces were more effective due to the formation of a diffusion layer, which contributed to
improving the fatigue strength of the steel specimens. The expected maximum pore size of the coatings treated in vacuum and electric
furnaces were estimated using statistics of extreme value.
 2007 Elsevier Ltd. All rights reserved.

Keywords: Thermal spray; Fatigue strength; Fracture mechanism; Fusing treatment; Adhesive strength; Statistics of extreme value

1. Introduction coating and substrate, are affected by the fusing treatment

process [4,5].
Thermal spray coatings, which consist of a lamellar In a previous study [3,6], we performed the post-fusion
structure of flattened and solidified fine particles, are por- treatment of a Co-based self-fluxing alloy coating onto
ous [1], and thus post-fusing treatment is necessary to medium steel applied using gas flame spray equipment in
change the phase composition of the coatings to decrease a vacuum furnace at temperature conditions of 930, 990,
the porosity and improve the fatigue strength [2,3]. Vac- 1050 and 1100 C, holding time of 0.5 h. As a result, the
uum and electric furnace treatments are used for this task, porosity and fatigue strength of thermal spray-coated spec-
but an induction heating system can also be applied for imens were significantly improved at temperature condition
large mechanical components or when precision tempera- of 1100 C. Also, we performed the post-fusion treatment
tures are required. The coating characteristics, and partic- to vary a holding time of 2, 4 and 7 h in a vacuum furnace
ularly the adhesive strength at the interface between the at temperature of 1100 C. Consequently, we observed a
40 lm diffusion layer at the interface of a Co-based self-
*
Corresponding author. Tel.: +82 55 279 7783; fax: +82 55 267 5142.
fluxing alloy coating onto medium steel after treating the
E-mail addresses: [email protected] (J. Oh), komotori@mech. specimen in a vacuum furnace at 1100 C for 4 h. A high
keio.ac.jp (J. Komotori), [email protected] (J. Song). adhesive strength of 170 MPa was measured using tensile

0142-1123/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijfatigue.2007.10.004
1442 J. Oh et al. / International Journal of Fatigue 30 (2008) 1441–1447

tests. A diffusion layer was also observed on specimens Table 1

treated in an electric furnace, and a correlation between Gas flame spraying parameters
the adhesive strength and diffusion layer thickness was Primary gas Oxygen (0.5 MPa)
deduced from the results. However, a diffusion layer did Secondary gas Acetylene (0.1 MPa)
Powder carrier gas Air
not form at the interface on specimens treated using an Spraying distance 150–200 mm
induction heating system. The much lower adhesive Rotational speed of specimen 300 rpm
strength of these specimens was mainly determined by the
mechanical anchoring effect between the coating and sub-
strate [7]. 2 h, held at 1100 C for 4 h, and subsequently cooled down
In this study, we altered the adhesive strength between in the furnaces. The I-series specimens were heated to
the coating and substrate by using three different fusing 1100 C for 120 s, held at 1100 C for 120 s, and then
treatment processes, making use of a vacuum furnace, an cooled down in air. After the fusing treatment was applied,
electric furnace, or an induction heating system. We then fatigue specimens were reproduced with a coating thickness
investigated the effect of the adhesive strength on the fati- of 1.0 mm. The surfaces of the fatigue specimens were
gue strength and fatigue fracture mechanism of fused spec- machined and then polished with emery paper and alumina
imens with a Co-based self-fluxing alloy on medium carbon powder (0.3 lm) until a mirror finish was achieved. Fatigue
steel substrates. tests were performed under rotating bending conditions at
3000 rpm. The fatigue fracture surfaces were then observed
2. Experimental procedure under a scanning electrical microscope (SEM) to study the
initiation sites of the fatigue cracks in detail as well as the
The substrate material was a medium carbon steel with a general morphology of the fatigue surfaces.
carbon content of 0.35%. After annealing the material at
1100 C for 3 h to produce a homogeneous substrate, it 3. Experimental results and discussion
was machined to the geometry and dimensions shown in
Fig. 1. The hashed region in the figure identifies the ther- 3.1. Characteristics of the thermal-sprayed coatings
mal-sprayed coating region. The surfaces were degreased
and grit-blasted prior to spraying, and the spray gun was Fig. 2 shows optical microscope images of the coating
set to give a thickness of 2.0 mm. A feedstock material con- microstructures before and after the fusing treatment.
sisting of Stellite-6 powder was applied to the hashed The ‘‘as-sprayed’’ coating shows irregular particles with
region of the specimens under atmospheric conditions lamellar microstructures (Fig. 2a), while the coatings of
using a gas flame spraying method. The chemical composi- the V-, A-, and I-series specimens show homogeneous
tion of the Stellite-6 powder was as follows (wt.%): 18.80 microstructures without lamellar structures (Fig. 2b–d),
Cr, 13.10 Ni, 7.68 W, 2.25 Fe, 2.28 Si, 1.78 B, 0.74 C, respectively. The I-series coating had the finest microstruc-
0.39 Mn, and balanced Co. Blasting was conducted at a tures due to the shortest fusing treatment time. The speci-
pressure of 0.5 MPa and a distance of 100 mm for 60 s with mens for as-sprayed, V-, A-, and I-series coatings had
alumina grit. The gas flame spraying conditions are shown porosities of 8.73, 0.23, 1.12, and 0.14%, respectively. The
in Table 1. After spraying, fusing treatments were per- porosity of the fused specimens was therefore significantly
formed using a vacuum furnace, an electric furnace, or less than that of the as-sprayed specimen. Fig. 3 compares
an induction heating system. Induction heat treatment is the Vickers hardness measured on the cross-sectional area
widely used for manufacturing automotive components of the fused specimens.
to improve fatigue and wear resistance. This process hard- Tensile test specimens, illustrated in Fig. 4, were pre-
ens only the surface layer of a material for a few seconds. pared to measure the mechanical properties of the fused
The advantages of this process include the accurate control coatings. These were produced by removing the substrate
of temperature, short processing time and less decarburiza- region, leaving only the fused coating layer. A 6 mm drill
tion [8,9]. The specimens were designated as V, A, or I-ser- was used to hollow out substrate region and a strain gauge
ies, respectively, according to the fusing treatment process. was attached to the narrowest diameter of the fused coat-
The V- and A-series specimens were heated to 1100 C for ing surface. The mechanical properties of the fused coating
and substrate-only specimens, shown in Table 2, were
much better than those of the as-sprayed coating
Co-based alloy specimens.
1.0 mm thick
3.2. Effect of the adhesive strength on the fatigue behavior
and fatigue strength
R2

105
0

210 Fig. 5 shows the rotating bending fatigue test results for
the V-, A-, I-series, and substrate-only specimens. The
Fig. 1. Typical geometry and dimensions of a fatigue specimen. nominal stress amplitude represents the maximum stress
 J. Oh et al. / International Journal of Fatigue 30 (2008) 1441–1447 1443

Fig. 2. Change in the microstructures of the (a) as-sprayed and post-fusing treated (b) V-series (c) A-series, and (d) I-series coatings.

800 Table 2
Mechanical properties of the coatings and substrate-only specimens
Coating Substrate
700
Specimen Young’s Tensile 0.2% proof Elongation
Vickers hardness, HV (2N)

600 modulus strength stress (MPa) (%)

(GPa) (MPa)
500 Substrate-only 205 580 254 22.0
As-sprayed 55 200 – –
400
V-series 250 900 803 0.45
300 A-series 230 780 670 0.42
I-series 240 750 650 0.40
V-series
200 A-series
I-series
100
0 0.5 1 1.5
Depth from surface (mm) 600
V-series
Fig. 3. Change in the Vickers hardness distribution of the fused coatings. 550 A-series
I-series
Substrate-only
500
Nominal stress amplitude (MPa)

450

400

350

300

250
Fig. 4. Hollow tensile test specimen.
200
applied at the coating surface. The fatigue strength of the
150
V- and A-series specimens was significantly higher than
that of the substrate-only specimens. However, the fatigue 100
strength of the I-series specimens did not increase as much 104 105 106 107 108
as that of the V- and A-series specimens. Number of cycles to failure
We examined the fatigue fracture mechanism of the Fig. 5. Relationship between nominal stress, ra, and number of cycles to
tested specimens in detail. Fig. 5 does not give the true failure, Nf, of the fatigue test specimens.
1444 J. Oh et al. / International Journal of Fatigue 30 (2008) 1441–1447
.X
stress levels applied at the fracture origin for the fused spec- r0i ¼ Ei My Ei I i ; ð1Þ
imens. Thus, the fatigue crack initiation sites were analyzed
for each series of specimens on the basis of the true stress where i = 1 is a coated layer and i = 2 is a substrate, M is
under a SEM. The fatigue cracks for all the V- and A-series the bending moment, I is the moment of inertia of the spec-
specimens originated at the pores in the coatings, which we imen area, and y is the distance from the central axis to the
termed ‘‘coated layer failure’’ fracture mode. Typical exam- fracture origin. The Young’s modulus, E, of the coated
ples are shown in Figs. 6b and 7a. The arrows in the images layer and the substrate materials are listed in Table 2.
show the crack initiation sites. Figs. 6d and 7c show a river The true stress amplitude, given in Fig. 9, indicates the
pattern, indicating a brittle fracture in the substrate region maximum stress applied at the fracture origin calculated
of the coated specimens during the final stage of the frac- by Eq. (1). The fatigue strengths for the V- and A-series
ture process. This suggests that the plastic constraint in specimens increased remarkably compared to the sub-
the substrate region of the coated specimens was affected strate-only specimens. In particular, the fatigue strength
significantly by the surrounding hard-coated layer [10]. of the V-series specimen was 2.8 times greater than that
As a result, it is difficult for the surrounding coating to of the substrate-only specimens. Thus, fused coatings with
form a minimal slip band in the substrate region due to a high durability caused a strong adhesive strength, which
the plastic constraint. For the I-series specimens, we contributed to effectively increase the fatigue strengths of
observed delamination of the entire coating from the sub- the V- and A-series specimens. The I-series specimens had
strate along the interface during the first stage of the frac- a much lower adhesive strength, leading to delamination
ture process due to the low adhesive strength, as shown in of the entire coating from the substrate along the interface
Fig. 8a. Several cracks were then initiated around the during the first stage of the fracture process. Although the
substrate region of the interface. We termed this an ‘‘inter- coating microstructures and mechanical properties of the I-
facial failure’’ fracture mode. Typical photographs of an series specimens improved over those of the as-sprayed
interfacial failure are presented in Fig. 8. The fatigue cracks specimens, no effective plastic constraint was displayed to
grew with time, producing striations that could be clearly prevent plastic deformation in the substrate region. As a
observed in typical ductile materials, as shown in Fig. 8c. result, their fatigue strength was almost equal to that of
The ductile fracture zone indicated in Fig. 8d, which was the substrate-only specimens. The fusing treatments in the
produced during the final stage of the fracture process, vacuum and electric furnaces effectively improved the
was located toward the center of the specimen. fatigue strengths of the thermal spray-coated steels.
We evaluated the true stress applied in the coating and We assume that residual stress in the coating for V- and
interface at the fracture origin using the following equation: A-series specimens might be almost relieved due to the

Fig. 6. Typical fracture surface features of the of V-series specimens (ra = 470 MPa, Nf = 3.74 · 106 cycles).
 J. Oh et al. / International Journal of Fatigue 30 (2008) 1441–1447 1445

initiated at the pores in the coating. In this section, we

develop a parameter to determine the fatigue strengths of
these specimens. Murakami et al. [12] demonstrated that
if the fatigue fracture origin is a small defect or a nonmetal
inclusion, the fatigue limit, rx, is determined by the matrix
Vickers hardness, HV, and the square root of the projected
pffiffiffiffiffiffiffiffiffi
area of the defects, area. They proposed fatigue limit pre-
diction equations taking the residual stress into consider-
ation, and showed that the fatigue limit p offfiffiffiffiffiffiffiffiffi
high strength
steels can be accurately predicted by this area parameter
model. pffiffiffiffiffiffiffiffiffi
The area parameter model is given as follows:

pffiffiffiffiffiffiffiffiffi 1=6 a
rx ¼ 1:56ðHV þ 120Þ=ð areaÞ ½ð1  RÞ=2 ;
a ¼ 0:226 þ HV  104 ; ð2Þ

where rx is a prediction of fatigue limit in MPa, p HV is

ffiffiffiffiffiffiffiffiffi
Vickers hardness of the coated layer in kg/mm2, area
is the square root of the projected area of the pores in
lm, and R is the stress ratio rmin/rmax. The scatter in
Vickers hardness distribution of Fig. 3 was within 10%.
Therefore, the Vickers hardness of the coated layer was
taken to be the average value.pFig. ffiffiffiffiffiffiffiffiffi 10 shows plots of
the cumulative probability of areamax in the coatings
of the V- and A-series specimens using extreme value sta-
tistics [12]. The standard inspection area was
S0 = 0.2 mm2, and the number of inspections was
pffiffiffiffiffiffiffiffiffi
n = 40. To determine areamax for N specimens from
Fig. 10, the return period T for one specimen was calcu-
lated by dividing the risk area S of a specimen by the
standard inspection area S0. The risk area of a specimen
with a diameter of 8 mm is S = pdl = p Æ 8 Æ 1 =
25.13 mm2 (the surface area region where the highest
Fig. 7. Typical fracture surface features of the A-series specimens stress is applied). The return periods were T = N Æ S/
(ra = 460 MPa, Nf = 1.94 · 106 cycles).
S0 = 628, 1260, and 12,600 for N = 5, 10, and 100 speci-
pffiffiffiffiffiffiffiffiffi
mens. The expected maximum sizes of the pores areamax
of the V-series specimens in Fig. 10 were 38.2, 41.2, and
post-fusion treatment in the furnace for a long time. Other 51.0 lm for N = 5, 10, and 100 specimens, p respectively.
ffiffiffiffiffiffiffiffiffi
investigator [5,11] reported that residual stress of the coat- The expected maximum sizes of the pores areamax of
ing was drastically reduced by an appropriate annealing the A-series specimens in Fig. 10 were 44.5, 47.5 and
treatment after thermal spraying. In case of the I-series 58.3 lm for N = 5, 10, and 100 specimens, respectively.
specimens, a slight compressive residual stress in the coat- Table 3 shows the measured size of pores at the fracture
ing might occur due to the post-fusion treatment for a origin and fatigue limit predicted by Eq. (2) for the V-
short time. However, the improvement effect on the fatigue and A-series specimens. The measured maximum sizes
strength of the I-series specimen was much lower compared of the pores at the fracture origin for these specimens,
to the V- and A-series specimens. Therefore, it is demon- as shown in Table3, were 37.9 and 43.7 lm, respectively.
strated that the effect of the adhesive strength on the Thus, the measured maximum size of the pores was al-
increase of fatigue strength was much higher than that of most equal to the expected maximum size. Based on
residual stress. above the discussion, the lower bound pffiffiffiffiffiffiffiffiffi of fatigue strength
was predicted using the value of areamax [13,14]. Fig. 11
3.3. Effect of the maximum size of the pores in the coating on shows the upper and lower bounds of the predicted fati-
the fatigue strength gue strength obtained using Eq. (2) for 10 and 100 spec-
pffiffiffiffiffiffiffiffiffi
imens, and the experimental results. The areamax used in
The V- and A-series specimens showed great fatigue Eq. (2) was obtained from Fig. 10. The lower bound of
strength improvements compared to the I-series specimens. the fatigue strength for the V- and A-series specimens
pffiffiffiffiffiffiffiffiffi
The fatigue cracks in all the V- and A-series specimens was well predicted by the areamax parameter.
1446 J. Oh et al. / International Journal of Fatigue 30 (2008) 1441–1447

Fig. 8. Typical fracture surface features of the I-series specimens (ra = 290 MPa, Nf = 1.75 · 106 cycles).

600 y
: Coated layer failure
F (%) Return period,
8 T = 1/(1-F)
550 : Interfacial failure 99.95
7 1200 T=628 for 5
99.9 specimens
500
6 99.8 500
Reduced variate, y = - ln ( - ln F)
True stress amplitude (MPa)

450 99.5 200

5
99 100
Cumulative Frequency

400 4
98 50
350 3 95 20

300 2 90

1 70
250 area max= 38.2 m
50
200 V-series
0 30
A-series
20 area max = 44.5 m
-1 10 V-series
150 I- series
1 A-series
Substrate-only
-2 0.1
100 0 10 20 30 40 50 60
104 10 5
106
10 7
10 8
area max ( m)
Number of cycles to failure
Fig. 10. Statistical distribution of the maximum size of the pores for the
Fig. 9. Relationship between the true stress, r 0 , and number of cycles to
V- and A-series specimens. S0 = 0.2 mm2, N = 40.
failure, Nf, of the fatigue test specimens.

coatings, as well as a fusing treatment to change the adhe-

4. Conclusions sive strength of the thermal-sprayed coatings, using three
different processes that utilized a vacuum furnace, an elec-
We performed rotational bending fatigue tests of ther- tric furnace, or an induction heating system. We then inves-
mal spray-coated steels with Co-based self-fluxing alloy tigated the effect of different adhesive strengths on the
 J. Oh et al. / International Journal of Fatigue 30 (2008) 1441–1447 1447

Table 3
Fatigue strengths applied at the fracture origin and fatigue limit predicted by Eq. (2) for the V- and A-series specimens
Series Nominal stress at Cycles to failure, Nf Average hardness, Pore size, Distance from r 0 (MPa) rx (MPa) r 0 /rx
pffiffiffiffiffiffiffiffiffi
surface, ra (MPa) HV (Kg/mm2) area ðlmÞ surface, h (lm)
V-series 470 3.74 · 106 440 31.6 379 498 491 1.01
480 2.17 · 106 440 33.7 247 509 486 1.05
490 5.45 · 105 440 37.9 426 520 477 1.09
500 2.39 · 105 440 26.2 23 530 507 1.05
520 1.65 · 105 440 30.5 304 551 494 1.12
A-series 450 3.56 · 106 395 38.2 64 466 438 1.06
460 1.94 · 106 395 40.2 25 476 434 1.10
470 6.46 · 105 395 43.7 50 487 428 1.14
480 2.16 · 105 395 35.3 32 497 444 1.12
r 0 : fatigue strength applied at the fracture origin using Eq. (1); rx: fatigue limit predicted by Eq. (2).

800 coated steels due to changes in the microstructures and the

V-series strong adhesion strength of the coating. All the fatigue
700 A-series Upper bound
cracks nucleated at the pores in the coatings so that the
= 1.6 HV fatigue strength was dominated by the maximum size of
(MPa)

the pores in the coatings.

600
Fatigue strength,

500
References
Number of specimens
for V- & A-series [1] Pawlowski Lech. The science and engineering of thermal spray
400 N = 10 Lower bound coatings. Tokyo: John Wiley and Sons; 1995.
[2] Stewart S, Ahmed R, Itsukaichi T. Rolling contact fatigue of post-
treated WC-NiCrBSi thermal spray coatings. Surf Coat Technol
300 N = 100 2005;190:171–89.
( )
1/ 6
= 1.56 ( HV + 120 ) / area ⋅ [(1 R) / 2] [3] Komotori Jun, Qu Ningshung, Shimizu Masao, Shirai Katsuyuki.
200 Fatigue properties of thermally sprayed steel with Co-based self-
200 250 300 350 400 fluxing alloy. Trans Jpn Soc Mech Eng 1996;62(604):2687–92.
Vickers hardness, HV [4] Hamatani Hideki, Miyazaki Yasunobu. Optimization of an electron
beam remelting of HVOF sprayed alloys and carbides. Surf Coat
Fig. 11. The predictions of the upper and lower bound of fatigue strength Technol 2002;154:176–81.
for 10 and 100 specimens and the experimental results. [5] Lesage J, Staia MH, Chicot D, Godoy C, De Miranda PEV. Effect of
thermal treatments on adhesive properties of a NiCr thermal sprayed
coating. Thin Solid Films 2000;377–378:681–6.
fatigue strength and fatigue fracture mechanism of the [6] Takeshita T. M.S thesis. Department of Mechanical Engineering,
Keio University; 1997.
specimens. For the specimens treated in vacuum and elec-
[7] Staia MH, Ramos E, Carrasquero A, Roman A, Lesage J, Chicot D,
tric furnaces at 1100 C for 4 h, the fatigue strength et al. Effect of substrate roughness induced by grit blasting upon
increased remarkably compared to the substrate-only spec- adhesion of WC-17% co thermal sprayed coatings. Thin Solid Films
imens. Thus, fused coatings caused a strong adhesive 2000;377:657–64.
strength due to the formation of a diffusion layer at the [8] Bertini L, Fontanari V. Fatigue behavior of induction hardened
notched components. Int J Fatigue 1999;21(6):611–7.
interface, which contributed to effectively increase the fati-
[9] Komotori J, Shimizu M, Misaka Y, Kawasaki K. Fatigue strength
gue strengths of the V- and A-series specimens. The I-series and fracture mechanism of steel modified by super-rapid induction
specimens had a much lower adhesive strength since a dif- heating and quenching. Int J Fatigue 2001;23:225–30.
fusion layer was not formed, although the microstructures [10] Kruzic JJ, McNaney JM, Cannon RM, Ritchie RO. Effect of plastic
and mechanical properties of the coating improved com- constraint on the cyclic and static fatigue behavior of metal/ceramic
layered structures. Mech Mater 2004;36:57–72.
pared to the as-sprayed specimens. This led to delamina-
[11] Yang Yung-Chin. Influence of residual stress on bonding strength of
tion of the entire coating along the interface during the the plasma-sprayed hydroxyapatite coating after the vacuum heat
first stage of the fracture process. Since no effective plastic treatment. Surf Coat Technol 2007;201(16):7187–93.
constraint was displayed to prevent plastic deformation in [12] Murakami Y, Takada M, Toriyama T. Super-long life tension-
the substrate region of the I-series specimens, their fatigue compression fatigue properties of quenched and tempered 0.46%
carbon steel. Int J Fatigue 1998;16(9):661–7.
strength was almost equal to that of the substrate-only
[13] Murakami Y. Metal fatigue: effects of small defects and non-metallic
specimens. inclusions. Tokyo: Yokendo Ltd.; 1993.
The vacuum and electric furnace fusing treatments effec- [14] Gumbel EJ. Statistics of extremes. NY: Columbia University Press;
tively improved the fatigue strengths of the thermal spray- 1957.