You are on page 1of 13

Effect of Hot-Isostatic-Pressing Parameters

on the Microstructure and Properties of Powder


Ti-6Al-4V Hot-Isostatically-Pressed Samples
K. ZHANG, J. MEI, N. WAIN, and X. WU

Ti-6Al-4V powders have been hot-isostatically-pressed (‘‘HIPped’’) using a range of


hot-isostatic-pressing (‘‘HIPping’’) conditions, and the effects on microstructure and mechanical
properties have been assessed. The properties were measured on test samples machined from
HIPped powder billets and on samples that contained the as-HIPped surface. The fatigue limit
of samples that contained the as-HIPped surface was improved by using a new HIPping pro-
cedure. The machined samples that had been HIPped at 1203 K (930 C) exhibited a better
balance of properties than those HIPped at 1153 K (880 C) or 1293 K (1020 C). The fine
microstructure, formed from the martensitic structure of the atomized powder, coarsens with
the increase of temperature or time during HIPping. These changes have been correlated with
the corresponding changes in properties and with the fracture surfaces. The significance of these
observations, especially the fatigue properties of samples that contain the as-HIPped surface, is
discussed in terms of the properties of net-shape HIPped components.

DOI: 10.1007/s11661-009-0149-y
 The Minerals, Metals & Materials Society and ASM International 2010

I. INTRODUCTION The surface finish is known to be the major factor that


influences the fatigue properties of a component, and
NET-SHAPE hot-isostatic-pressing (‘‘HIPping’’) is net-shape hot-isostatically-pressed (‘‘HIPped’’) products
becoming an accepted technology for the manufacture are designed to retain much of the as-HIPped surface.
of a range of engineering components.[1–4] The potential Thus, the influence of the ‘‘as-HIPped surface’’ on the
economic advantages of net-shape HIPping of powder fatigue properties of a component produced using net-
to produce complex-shaped and fully dense components shape HIPping is an important aspect of the work
from titanium and nickel-based alloys[5–7] and other reported here. The term as-HIPped surface is used to
expensive metal powders have been investigated. In this describe the final surface of the component after
process, a containing capsule, together with internal HIPping after the tooling has been removed. This
tooling, is filled with powder; the geometry of the surface is in direct contact with the mild steel tooling
capsule and tooling is defined via computer modeling of during HIPping, and the as-HIPped surface is generated
the shrinkage of the powder compact during the after HIPping by acid etching to remove the tooling.
densification process occurring during HIPping. The The overall aim of the work reported in this article is
capsule is pumped to backing pump pressure and sealed to study the effect of HIPping variables (HIPping
vacuum tight, prior to HIPping under defined condi- temperature and HIPping procedure) on microstructure,
tions. The tooling and capsule (commonly of mild steel) surface finish, and mechanical properties (tensile,
are then removed by machining or pickling.[7] Net-shape fatigue, and fracture toughness) of HIPped Ti-6Al-4V
HIPping can be used with titanium-based materials to powder; the main interest is the properties of samples
obtain cost reduction by reducing the amount of difficult that contain the ‘‘as-HIPped’’ surface. There is a wealth
titanium machining and the large amount of material of data on properties of conventionally processed
loss.[8] It is essential that the factors that influence the samples (e.g., References 9 through 13) with which
properties of such components are understood and that these data can be compared. In earlier work, it was
the HIPping process parameters are optimized accord- shown, using conventional HIPping, that the surface
ing to specific application requirements. roughness downgraded the fatigue properties of
as-HIPped samples.[14] Thus, during conventional
HIPping, when the temperature and pressure are ramped
up in phase, the steel yields so that a rough surface on the
K. ZHANG, formerly Research Fellow, with the IRC in Materials,
The University of Birmingham, is Metallurgist, with Intertek Derby, scale of the powder particles is generated. This rough-
ELT14 Rolls Royce, Elton Road, Derby, Derbyshire, DE24 8BJ, ness is caused by the penetration of the powder particles
United Kingdom. J. MEI and N. WAIN, Research Fellows, and into the mild steel tooling, which occurs because the
X. WU, Professor, are with the IRC in Materials, The University of steel has a lower yield stress at low temperatures than
Birmingham, Edgbaston B15 2TT, United Kingdom. Contact e-mail:
x.wu.1@bham.ac.uk does Ti-6Al-4V. As part of the current study, a new
Manuscript submitted August 20, 2009. HIPping procedure is used, which eliminates this surface
Article published online February 25, 2010 roughness and thus improves the fatigue properties.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1033


II. EXPERIMENTAL METHODS HIPped at 1203 K (930 C). Thus, using the new
AND MATERIALS procedure, the powder particles are flattened by, rather
than penetrating, the tooling, as occurs when the
The Ti-6Al-4V powder in the size range of 50 to temperature and pressure are increased in phase.[14]
400 lm used in this work was produced by StarMet With this technique, it takes about 2 hours to reach the
(Concord, MA), using the plasma rotating electrode selected pressure, during which time the samples are at
process (PREP). The as-received powder was generally the HIPping temperature. The samples are then held for
HIPped at 100 MPa for 4 hours at a preselected tem- a further 3 hours at the HIPping temperature, under the
perature using a fixed heating rate of 5 K/min (5 C/min). HIPping pressure, and subsequently cooled down. The
Chemical analyses of HIPped samples were carried out powder is thus held at the HIPping temperature for
by Incotest (Hereford, UK), and the analysis of the 5 hours, i.e., for 1 hour longer using the new procedure
samples for which the data are reported in this article is than during a conventional HIPping cycle.
shown in Table I. Measurements of the beta transus Some samples were partially HIPped, by reducing the
showed that it was about 1293 K (1020 C), as expected HIP temperature to 1123 K (850 C) and cooling the
for Ti-6Al-4V of this composition; the calculated transus sample as soon as it had reached 1123 K (850 C).
is 1285 K (1012 C) (S. Fox, stephen.fox@timet.com, In this case, the temperature and pressure were ramped
private communication). In one HIPped sample, the in phase. These experiments were carried out so that the
O content was found to be 2300 ppm rather than between microstructural and morphological changes that occur
1700 and 2000 ppm, as found in all other samples. during HIPping could be followed at the early stages of
The HIP cans, which were made from A1018 mild consolidation. These observations were compared both
steel with a surface finish of 1.6 lm, were typically with conventionally HIPped samples and with observa-
45 mm in diameter and 140-mm long, which allowed tions made on individual powder particles that were
tensile and fatigue samples to be machined from the simply heated to the HIPping temperature and held
same HIPped sample. The conventional HIP procedure there for 2 hours, without the application of pressure.
is to ramp temperature and pressure in phase, taking Three cylindrical M8 tensile samples were machined
about 3 hours to reach the selected HIP temperature, from HIPped compacts for each HIP condition in order
after which it is held, usually for 4 hours, before cooling
at about 5 K/min (5 C/min). Some samples were
HIPped using a different HIP procedure (U.K. Patent
Application 0807703) in order to improve the surface
finish of the as-HIPped samples. In this procedure, the
capsule with powder is heated to the HIPping temper-
ature, again taking about 3 hours, before the pressure is
increased. The basis of this patent is illustrated by the
compression tests illustrated in Figure 1. These tests
were carried out at the temperatures indicated on
samples of powder Ti64, which had been previously
HIPped at those temperatures. In addition, samples of
fully HIPped Ti64 and of the mild steel used as canning
material were tested at these temperatures. The data
show that the tooling is stronger than the partially
HIPped Ti-6Al-4V HIPped and tested below about
973 K (700 C). At these low temperatures, HIPping
results in samples that are only about 70 pct of full
density, as indicated on the curve for the Ti64 powder.
Between about 973 K (700 C) and 1123 K (850 C), as
consolidation progresses to between 80 and 90 pct full
density, the partially HIPped Ti64 is stronger. At higher Fig. 1—Graph showing the temperature dependence during HIPping
temperatures, the steel is stronger and in fact, at 1203 K of the 0.2 pct proof stress of powder Ti64, together with the flow
(930 C), the standard HIP temperature, it is about stress of the mild steel tooling and of fully HIPped Ti64. The macro-
twice as strong as the Ti64 even though the Ti64 HIPped scopic density of the partially HIPped powder is seen to increase
at this temperature is now >99.5 pct density. At this from about 70 pct full density at 873 K (600 C) to >99.9 pct at
temperatures above 1173 K (900 C), which results in an initial in-
temperature, the 0.2 pct proof stresses of the two crease in proof stress, followed by the expected decrease at high tem-
samples of Ti64 are identical, as would be expected peratures. Note that the individual partially HIPped samples are
since prior to testing these samples had both been tested at the temperature at which they have been HIPped.

Table I. Chemical Composition (Weight Percent) of Ti-6Al-4V Powder HIPped at 1203 K (930 °C)

Alloy O C H N Fe Al V Ti
HIPped Ti-6Al-4V 0.20 0.034 0.0039 0.006 0.17 6.42 3.88 balance

1034—VOLUME 41A, APRIL 2010 METALLURGICAL AND MATERIALS TRANSACTIONS A


to assess the tensile properties of HIPped samples (i.e., A JEOL* 7000 scanning electron microscope (SEM)
of samples from which the as-HIPped surface had been
removed). One sample, which was HIPped using the new
HIP procedure and which contained the as-HIPped *JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.
surface, was tested in tension. The complex method
required to manufacture such a sample has been was used to examine the powder particles’ morphology
described earlier.[14] In this method, the gage length of and cross sections and the solid specimens’ cross
the sample was produced to net shape but the rest was sections. The fracture surfaces of failed tensile and
produced to near net shape, at successively increasing fatigue samples were examined using a PHILIPS**
diameters (in a series of steps). Final machining was
done on the near net-shape sections to produce the
conventionally tapered section and threaded ends of the **PHILIPS is a trademark of FEI Company, Hillsboro, OR.
tensile sample. This method was used because it is very
difficult to produce the tapered geometry of a tensile
XL30 SEM or JEOL 6060 SEM. The Inca (Oxford
sample directly by HIPping, but it is straightforward to
Instruments, Abingdon, Oxfordshire, UK) electron
produce to net shape the parallel-sided cylindrical gage
backscattered diffraction (EBSD) system was an integral
length and the stepped contour at either end.
part of the JEOL 7000 SEM, which was used to provide
To investigate the influence of the as-HIPped surface
information about the crystallographic orientation in
on the fatigue properties, flat HIPped plates 120 mm in
original powders and in HIPped samples. The operating
length, 60 mm in width, and 10 mm in thickness were
parameters were as follows: specimen tilt angle of
cut by EDM along their length, after removing the mild
70 deg, accelerating voltage of 20 kV, and working
steel can by acid pickling in 20 pct nitric acid. The EDM
distance of 12 to 16 mm. The step size for the mapping
cut surfaces were final machined by reciprocating
was 0.4 lm, the pixel area was 0.05 lm2, and the low-
surface grinding (RSG) to generate the side surfaces of
angle boundaries (<5 deg) were not counted.
the samples, and the original plate surfaces, which had
Hardness tests using a Vickers hardness with a 20-kg
been in contact with the mild steel can during HIPping,
load were carried out on the as-HIPped surface and on
became the top and bottom surfaces of the test pieces.
cross sections of HIPped samples. In addition, taper
Some of these specimens with as-HIPped surfaces were
sections were prepared from HIPped samples so that the
electropolished (using 60 vol pct Methanol, 35 vol pct
rate of change of hardness with distance from the
1-Butanol, and 5 vol pct HClO4, at a temperature of
as-HIPped surface could be assessed. The taper sections
238 K (–35 C) and a current of 0.8 to 1 A) for 15 to
were spark machined from the as-HIPped sample and
20 minutes so that fatigue data could be obtained for the
were lightly hand polished to 1200 grit. These polished
bulk sample without any influence of surface machining.
taper sections were examined using the JEOL 7000 SEM
Compact tension specimens were machined from
to assess the microstructure and to determine if any
HIPped bars to conduct fracture toughness tests accord-
significant changes in composition were detectable as a
ing to the relevant British Standard.[15]
function of distance from the as-HIPped surface.
Tensile tests were carried out using a Zwick
(Leominster, UK) testing machine at an initial strain
rate of 10 3 s 1 in air at room temperature. Three III. RESULTS
specimens were used for each HIPping condition in
order to obtain some indication of the repeatability of X-ray diffraction patterns show that the powder is
the properties. Four-point bending fatigue tests were martensitic, as would be expected from the rapid cooling
carried out using an Amsler Vibrophore electromagnetic rates associated with the production of PREP
resonance testing machine for fatigue life evaluation. powder,[16,17] between about 103 and 104 K s 1 (103 and
Typically, ten specimens were used to determine the 104 C s 1). The spherical morphology of the as-received
fatigue limit. Tests were conducted at room tempera- powder is shown in Figures 2(a) and (b). Cross sections
ture, at a stress ratio R of 0.1 and a frequency of ~95 Hz. of these powder particles (e.g., Figure 2(c)) confirm that
A stress that corresponded to a life beyond 1.0 9 107 they are composed entirely of martensitic alpha-prime
cycles was defined as the fatigue limit, i.e., the runout of phase. The powders are mostly polycrystalline, as seen
the specimen. in Figure 2(d).
For fracture toughness testing, specimens were
precracked at room temperature, using an Amsler
Vibrophore (Stratford upon Avon, UK) testing A. Effect of HIPing Temperature on Microstructure
machine, via fatigue to give starter crack lengths within and Mechanical Properties
the range set out in the standard. The maximum applied 1. Microstructure of samples HIPped using different
loads during precracking were reduced from 8 to 2 kN temperatures
during cycling; the stress ratio was 0.1 and the frequency The effect of HIPping temperature was investigated
was 60 to 70 Hz. The final fracture toughness testing using a range of temperatures, and Figure 3 shows the
was carried out on a servohydraulic testing machine microstructure and EBSD maps of samples HIPped at
manufactured by ESH Ltd (Telford, Shropshire, UK) 1153 K (880 C), 1203 K (930 C), 1293 K (1020 C),
with a ramp rate of 0.5 mm/min to obtain load vs crack and 1313 K (1040 C). These conventionally HIPped
opening displacement curves. samples were found to be fully dense, as judged by the

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1035


Fig. 2—(a) and (b) SEM SE micrographs showing the morphology of as-received PREP Ti-6Al-4V powders; (c) SEM BSE micrograph of cross
section of the powder showing martensitic alpha-prime phase; and (d) EBSD map showing orientations of the martensite laths.

absence of any visible porosity in conventional SEM only partially HIPped. In Figure 4(a), taken from a
images, and in the case of samples HIPped at 1203 K powder particle heat treated at 1203 K (930 C) for
(930 C), tomography showed that the porosity was less 2 hours, it is clear that the martensitic structure of the
than 0.02 pct. The as-HIPped microstructures, obtained original powder particle has transformed to alpha and
for samples HIPped below the transus, are composed of beta, effectively on the same scale as the original alpha
equiaxed grains, which appear to have formed both at prime martensite. No equiaxed grains are obvious at the
the former powder particle boundaries and at grain prior particle surface, although some may be present
boundaries inside the powders, together with a mainly within the original powder particle. In contrast, in the
lathlike structure in the central part of the original partially HIPped samples shown in Figures 4(b) and (c),
particles. The sample HIPped above the transus at the development of equaixed microstructure is obvious
1313 K (1040 C) shows a beta-transformed microstruc- where particles are in contact. Hence, it is suggested that
ture, with alpha formation strongly influenced by the the equiaxed phases in fully HIPped samples (Figure 3)
beta boundaries, as would be expected.[18,19] This are formed at prior particle boundaries (PPBs) during
microstructure is obviously very different from those HIPping, because sufficient deformation occurs locally
of samples HIPped below the transus to cause recystallization. The limited amount of equi-
In order to understand the as-HIPped microstructure, axed grains formed within particles are formed at prior
powders were either heat treated at a selected subtransus beta boundaries, perhaps because diffusion is faster in
HIPping temperature (with no pressure applied) or were these regions than within grains.

1036—VOLUME 41A, APRIL 2010 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 3—SEM BSE micrographs in which the dark phase is a and EBSD map of Ti-6Al-4V samples HIPped at (a) and (b) 1153 K (880 C),
(c) and (d) 1203 K (930 C), (e) and (f) 1293 K (1020 C), and (g) BSE micrograph of sample HIPped at 1313 K (1040 C) showing
(transformed microstructure. (h) EBSD map of b phase of sample HIPped at 1293 K (1020 C) (cf. (f)).

The microstructures simply coarsen as the tempera- very close to the b transus, the microstructure coarsened
ture is increased below the b transus. When the HIPping significantly, as shown in Figures 3(e) and (f), but still
temperature was increased to 1293 K (1020 C), i.e., retained the same microstructural features of laths and

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1037


Fig. 3—Continued.

equiaxed grains. When the sample was HIPped at 0.2 pct proof stress was slightly higher (900 MPa) than
1313 K (1040 C), a b transformed microstructure was that found for a similarly HIPped sample (886 MPa),
obtained, as shown in Figure 3(g). which had the HIPped surface machined off, and the
It should be noted that these EBSD maps in Figure 3 elongation was further reduced from 12 pct to about 8 pct.
and elsewhere show the orientation of the alpha phase
and the limited volume of beta appears as unindexed 3. Fatigue properties and observations of fracture
volumes. An example of a beta EBSD is shown in surfaces of fatigued samples machined from HIPped bars
Figure 3(h), where it is clear that the volume fraction of Figure 6 shows a plot of stress amplitude vs number of
the beta is small. These regions will be ignored in the cycles to failure (SN curve) for the samples machined
following description of the microstructure. from bars HIPped at different temperatures. The samples
HIPped at 1203 K (930 C) exhibited the highest fatigue
2. Tensile properties and examination of fracture limit of 500 MPa (at 1.5 9 107 cycles). The fatigue limit
surfaces of specimens HIPped at 1153 K (880 C) was 400 MPa,
A summary of the tensile data obtained from the whereas for samples HIPped at 1293 K (1020 C), the
HIPped samples is given in Table II. Samples HIPped at fatigue limit was decreased further to 300 MPa.
1153 K (880 C) had the highest 0.2 pct proof stress and The SEM micrographs of fracture surfaces near crack
a relatively low elongation of about 10 pct. Samples initiation sites are shown in Figure 7. It can be seen that
HIPped at 1293 K (1020 C) were shown to have the fatigue cracks originated from the surface in all speci-
lowest proof stress and tensile strength, and an elonga- mens. In the fractured surfaces, river patterns could be
tion similar to that of samples HIPped at 1153 K seen. The main mode of fatigue failure of the samples is
(880 C). The samples HIPped at 1203 K (930 C) had transgranular. For samples HIPped at 1153 K (880 C)
proof stresses intermediate between those of samples and 1203 K (930 C), groups of transgranular facets
HIPped at 1153 K (880 C) and 1293 K (1020 C) but can be identified as the fatigue crack initiation sites
had the largest elongation up to 21 pct. (Figures 7 (a) and (b)). This form of faceting is also
Fracture surfaces of the tensile specimens were typical of the early stages of fatigue fractures in wrought
examined, as shown in Figure 5. Figure 5(a) shows the forms of Ti-6Al-4V, in other a/b alloys, and in near a
fracture surface of a specimen HIPped at 1153 K titanium alloys in general.[20]
(880 C), where groups of transgranular facets can be For samples HIPped at 1293 K (1020 C), large facets
seen. The crack appears to have propagated along a PPB are seen, as shown in Figure 7(c). These large facets can
(indicated by the arrow in Figure 5(a)). There were no be formed[21] when the crack extends through a packet
obvious crack initiation sites in samples HIPped at of aligned alpha laths (indicated by the arrow) that have
1203 K (930 C); a typical ductile fracture surface can similar orientations.
be seen in Figure 5(b). In the specimens HIPped at
1293 K (1020 C), cracking was characterized by flat 4. Fracture toughness of HIPped samples
transgranular facets (Figure 5(c)), which were on the Room-temperature fracture toughness testing was
scale of the coarser lamellar colonies in this sample conducted to further assess the properties of powder
(Figures 3(e) and (f)). Ti-6Al-4V samples HIPped at 1203 K (930 C) and
The tensile properties of a sample that contained the 1293 K (1020 C). Values of fracture toughness are
as-HIPped surface made using the new HIPping proce- shown in Table III, which shows that there is no
dure were also measured, and it was found that the significant difference.

1038—VOLUME 41A, APRIL 2010 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 4—SEM BSE images of PREP powder of Ti-6Al-4V showing (a) an isolated powder particle that has been held for 2 h at 1203 K (930 C)
under ambient pressure (no equaixed region is obvious) and (b) and (c) partially HIPped samples where the development of equiaxed structure at
PPBs is apparent (as indicated by the arrow).

Table II. Room-Temperature Tensile Properties for Powder Ti-6Al-4V HIPped at Different Temperatures; Results Shown are for
the Three Samples Which Were Tested for Each Condition; Typical Data for Forged Samples are Also Shown as a Comparison

Processing Conditions 0.2 Pct Proof Stress (MPa) UTS (MPa) Elongation (Pct)
HIPping at 1153 K (880 C)/100 MPa/4 h 925, 927, 931 989, 994, 998 9, 10,11
HIPping at 1203 K (930 C)/100 MPa/4 h 899, 901 904 970, 971, 972 18, 19, 21
HIPping at 1293 K (1020 C)/100 MPa/4 h 836, 839, 840 913, 913, 913 10, 11, 12
Sample forged at 1193 K (920 C) to 1213 K (940 C)[19] 875 965 15

The best balance of properties (tensile, fatigue, and B. Effect of HIP Schedule and of Post HIP Surface
fracture toughness) was obtained for samples HIPped at Treatments on Mechanical Properties
1203 K (930 C), and this temperature was therefore 1. Nature of as-HIPped surfaces
chosen as the HIPping temperature for subsequent tests Figure 8 shows secondary electron micrographs of
where the effects of surface finish and of the influence cross sections of HIPped Ti-6Al-4V surfaces. It is clear
of the as-HIPped surface on fatigue properties were that the as-HIPped surface, made using the conventional
investigated.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1039


Fig. 5—SEM SE micrographs showing the fracture surfaces of tensile failed samples HIPped at (a) 1153 K (880 C) (crack propagating along a
PPB is indicated by the arrow), (b) 1203 K (930 C), and (c) 1293 K (1020 C).

surface and the interior and that the hardness change


extends to a depth of about 200 lm into the sample, as
indicated in Figure 8(a). The WDX analysis showed
(Figure 9(b)) the rate of change of C count measured on
taper sections, which follows the hardness change.
Metallographic examination and image analysis of the
taper sections showed a small increase in alpha content
at the surface of the as-HIPped sample, as shown in
Figure 9(c).

2. Influence of HIPping procedure on the tensile


properties of HIPped samples
Fig. 6—SN fatigue curves for powder Ti-6Al-4V samples with Test samples, which were machined from HIPped
machined surface HIPped at different temperatures showing that the
fatigue limit of samples HIPped at 1203 K (930 C) was higher billets made using the new HIPping procedure, had a
than those obtained at 1153 K (880 C) and 1293 K (1020 C). The proof stress of 886 MPa and an elongation of 12 pct.
arrows indicate that the tests were interrupted without the failure of These properties are somewhat lower than those
the samples. obtained using the conventional HIPping process shown
in Table II (yield of 901 MPa and elongation of 19 pct)
procedure (Figure 8(a)), is quite rough with the rough- and are associated with the slight coarsening of the
ness being on the scale of the original powder particle microstructure caused by the longer time at the HIP
size.[14] A much smoother surface is obtained using the temperature (compare Figures 8(c) and (d)) even though
new procedure, as shown in Figure 8(b). oxygen content of this HIPped sample (2300 ppm) is
Figure 9(a) compares the surface and interior hard- higher. The fracture surfaces of failed tensile sample
ness of a sample HIPped using the new procedure and HIPped at 1203 K (930 C) using the new procedure
also shows the rate of change of hardness from the show transgranular facets formed by clusters of lamellae
as-HIPped surface measured on taper sections cut at an with similar orientations around the crack initiation site
angle of 2.5 deg. It is clear that there is a small, but (Figure 10). The properties lie between those of samples
significant hardness difference between the as-HIPped conventionally HIPped at 1203 K (930 C) and 1293 K

1040—VOLUME 41A, APRIL 2010 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 7—Secondary electron SEM micrographs of fracture surfaces of fatigue failed samples HIPped at (a) 1153 K (880 C), (b) 1203 K (930 C),
and (c) 1293 K (1020 C). The fatigue failure of the samples is associated with transgranular facets.

Table III. Room-Temperature Fracture Toughness Values highest roughness with Ra = 9.1 lm, as noted
Obtained from Powder Ti-6Al-4V Samples HIPped earlier,[14] and the machined surface had the lowest with
at Different Temperatures Ra = 0.4 lm. The roughness values of the electropo-
lished surface and of the as-HIPped surface made using
Fracture
Toughness, the new HIPping procedure were between these two
Specimen KQ (MPaÆm) extremes.
The SN curves for HIPped Ti-6Al-4Vsamples with the
HIPped Ti-6Al-4V at 1203 K (930 C) 71/74/76/77 four different surface finishes are shown in Figure 11.
HIPped Ti-6Al-4V at 1293 K (1020 C) 77/80 Samples with as-HIPped surfaces made using the new
HIPping procedure had the highest fatigue limit
(at N > 107 cycles) of 620 MPa. The fatigue limit
(1020 C), as would be expected from the microstructure of samples with as-HIPped surfaces made using the
that is intermediate between these. conventional HIPping procedure[14] was confirmed as
much lower, at 350 MPa. Conventionally HIPped
3. Influence of surface finish on the fatigue strength samples that were machined or electropolished after
of HIPped samples HIPping have fatigue limits between these two.
Four sets of samples with different surface finishes The fracture surfaces of samples failed in fatigue were
were produced to assess the influence on the fatigue examined to investigate the crack initiation characteris-
properties for samples HIPped at 1203 K (930 C). The tics and the crack initiation; all are associated with
first three types of sample were made using the conven- transgranular facets consisting of a cluster of a and b
tional HIPping procedure and were samples (a) with the laths with a similar orientation, as seen in Figure 7(b).
as-HIPped surface; (b) HIPped and electropolished, as
described in the experimental section; (c) HIPped and
machined by RSG; and (d) containing the as-HIPped IV. DISCUSSION
surface made using the new HIPping cycle. Table IV
shows the surface roughness of the four surface condi- There are two aspects of the experimental results
tions. It is obvious that the as-HIPped surface made obtained in this work on HIPping of Ti-6Al-4V, which
using the conventional HIPping procedure had the will be the main focus of this discussion. These aspects

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1041


Fig. 8—Secondary electron SEM images of cross sections of HIPped Ti-6Al-4V surface: (a) rough surface obtained using conventional HIPping
procedure and (b) smooth surface obtained using the new HIPping procedure. SEM BSE images of PREP powder of Ti-6Al-4V HIPped using
(c) the conventional procedure and (d) the new procedure showing coarser microstructure in the HIPped sample.

are (1) the influence of the nature of the surface on the into blunt undulations and, thus, result in a decrease in
fatigue and tensile properties of samples and (2) the the magnitude of the stress concentration. The machined
influence of HIPping temperature and HIPping proce- surface has a low roughness, but some surface damage
dure on the tensile and fatigue properties of HIPped was produced during the machining process.
samples where any influence of the HIPped surface is In the samples HIPped using the new procedure,
removed by machining samples from larger HIPped flatter surfaces are formed, because no longer do the
samples. These two aspects are discussed, in turn, before particles penetrate the steel to form sharp features. It
discussing the factors that influence the microstructure should be noted that the scatter observed in these
of the HIPped samples. samples is larger than that found in the samples that
have been HIPped conventionally. It is suggested that
this larger scatter could be associated with the presence
A. Influence of the Nature of the Surface of As-HIPped
of some remnant surface dimples; samples with grooves
Samples on Fatigue and Tensile Properties
will fail earlier than those without grooves. Further
The fatigue response of samples HIPped using the work on optimizing surface quality is thus of interest.
conventional HIPping schedule, and the response of The combined effect of low surface roughness and the
samples HIPped using the new procedure, which devel- absence of any significant surface damage can lead to
ops a smoother surface, show the importance of surface the improved fatigue life of the sample.[14] This is an
finish on fatigue life, which is of course well docu- important observation because it means that compo-
mented. In the HCF regime, the total fatigue life nents made by net-shape HIPping do not require
depends principally on fatigue crack initiation and early any surface machining, and indeed the fatigue limit
growth behavior.[22] The sharp grooves formed by of 620 MPa obtained with samples containing the
powder particles on the conventional as-HIPped surface as-HIPped surface obtained when using the new HIP
serve as sites of stress concentration and promote crack schedule is within the range of those reported for
initiation, and the reproducibility of these data probably wrought samples.[23–25]
reflects the fact that these stress concentrations are In the earlier work on conventionally HIPped
present on every sample. Electropolishing removes samples,[14] the elongation was slightly degraded from
about 15 to 20 lm, and the sharp grooves were turned about 21 pct for samples machined from the HIPped

1042—VOLUME 41A, APRIL 2010 METALLURGICAL AND MATERIALS TRANSACTIONS A


earlier work, but it is still only a small, although
significant decrease. This slight reduction in elongation
and the significant improvement in fatigue properties
are associated with the smooth as-HIPped surface
(Figure 8) and with the increase in hardness (Figure 9(a))
observed in the near-surface region of the as-HIPped
sample, but the relative importance of these two
independent factors in increasing the fatigue limit is
not clear. This hardness increase is coupled with a
detectable increase in the carbon counts near the surface
(Figure 9(b)), which would be expected since diffusion
of carbon from the mild steel casing would occur during
HIPping. The activation energy for diffusion of carbon
into beta titanium is reported as between 150 and
200 kJ/mol[26,27] and diffusion constant (Do) at 1223 K
(950 C) as 108 cm2/s,[27] which would lead to a value of
Dt during the 3 hours of HIPping at 1203 K (930 C)
within a factor of 2 of 500 lm. This calculation, which is
of course approximate, has ignored diffusion through
alpha. The observed increase in carbon (an alpha
stabilizer) counts also results in a small increase in the
proportion of alpha phase near the surface (Figure 9(c)),
which may also account for the slight reduction in
elongation observed. According to the measurements in
Figure 9, the hardness and microstructural changes
extend to a depth of about 200 lm, which is compatible
with the distance over which the change in C count is
observed. Increase of surface carbon content and
consequent hardness can increase the fatigue strength
as long as no obvious alpha case forms. Apparently, in
the present study, the 1 to 2 pct increase in alpha volume
fraction does not form a deleterious alpha case. It would
be of interest to carry out a study where the diffusion of
C could be significantly reduced using the low C iron
(0.01 wt pct C) as a tooling material,[28] but it would be
necessary to assess its strength at the HIPping temper-
ature since if it is softer than the Ti64 indentations of the
sort shown in Figure 8 could form. A comparison of
fatigue limit between the as-HIPped samples obtained
using mild steel and ultralow C steel under the new
HIPping procedure will clarify whether the smooth
surface or diffusion of C is the key for improving the
Fig. 9—(a) Values of Vickers hardness, (b) carbon counts obtained
by WDX analysis, and (c) fraction of alpha taken from taper fatigue property.
sections of a sample of Ti-6Al-4V HIPped using the new HIPping
procedure. The x-axis is the approximate distance from the as-HIPped
surface obtained from the measured positions of the data points. B. Influence of HIPping Temperature and HIPping
Values of hardness obtained from measurements on the as-HIPped Procedure on the Tensile and Fatigue Properties
surface (S) and on a cross section (X) are also indicated in (a). of Machined Samples
The fracture surfaces of the tensile specimens in the
samples to 19 pct for samples that were tested with the specimens HIPped at 1153 K (880 C) show that the
as-HIPped surface present. Thus, in the earlier work, the crack appears to have propagated along a PPB, sug-
presence of the as-HIPped surface was found to decrease gesting that inadequate bonding occurs at the HIPping
the ductility slightly. In the present work, the lower temperature of 1153 K (880 C). In the specimens
value of elongation of 12 pct, found in samples HIPped HIPped at 1293 K (1020 C), cracking was character-
using the new HIPping procedure and that were ized by flat transgranular facets that were on the scale of
machined from the bulk, is associated with a signifi- the coarser lamellar colonies in this sample and the
cantly higher oxygen content of 2300 ppm compared corresponding clusters of these coarser regions. It thus
with 1700 to 2000 ppm in the earlier work and coarser appears that the temperature of 1203 K (930 C) is an
microstructure in the samples HIPped using the new optimum between the limited bonding at 1153 K
HIPping procedure. The further decrease in elongation (880 C) and the coarsening of the microstructure at
to 8 pct, found in samples that contained the as-HIPped 1293 K (1020 C). The tensile properties of samples
surface, is slightly larger than the decrease found in the HIPped at 1203 K (930 C) using the new HIPping

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1043


Fig. 10—SEM secondary electron micrographs showing the fracture surfaces of failed tensile sample HIPped at 1203 K (930 C) using the new
procedure. Transgranular facets formed by clusters of lamellae can be seen in the crack initiation site (elongation = 13 pct).

shown that fatigue cracks initiate from clusters of a laths


that have similar grain orientation. Such clusters could
fail as a large individual grain. For samples HIPped at
1293 K (1020 C), a cluster of coarser alpha laths, which
have similar crystallographic orientations (Figures 3(e)
and (f)), can be formed, with fine beta laths between the
coarse alpha laths. Such a large size cluster can result in
earlier crack initiation and rapid crack propagation. For
samples HIPped at 1153 K (880 C), the beta grains and
laths are fine, which cannot deflect or blunt a crack, so
the crack will cut through the beta laths and propagate
more easily than through a coarser structure.[26]
In addition, the imperfect bonding illustrated in
Figure 5(a) will also reduce the fatigue life if a crack
Fig. 11—SN fatigue curves for HIPped Ti-6Al-4V samples with four
different surface finishes. Samples with the as-HIPped surfaces nucleates close to a region where bonding is inadequate.
formed using the new HIPping procedure had the highest fatigue The sample HIPped at 1203 K (930 C) has a coarser
limit. The arrows indicate that the tests were interrupted without the microstructure than the samples HIPped at 1153 K
failure of the samples. (880 C), but finer than that those HIPped at 1293 K
(1020 C), and it appears that this structure offers an
procedure are intermediate between the properties of the optimum balance for fatigue when using conventional
samples conventionally HIPped at 1203 K (930 C) and HIPping.
1293 K (1020 C), which is reasonable in view of the The increase of fracture toughness with the increase of
scale of the microstructure, which is intermediate HIPping temperature from 1203 K (930 C) to 1293 K
between these samples. In view of this observation, it (1020 C) is likely to be associated with the coarsened
may be that the tensile properties could be improved if a HIPped microstructure, which can lead to more marked
slightly lower HIPping temperature were used when deflection of cracks.[26]
using the new HIP procedure, and this may lead to a The overall scale of the microstructure in the
further improvement in fatigue properties. as-HIPped samples clearly contributes to the balance
The fatigue response of the machined samples, which of properties, and the origin of this fine microstructure is
were HIPped at three different temperatures using the the fact that the microstructure of the as-received
conventional HIPping schedule, shows that samples powder particles is martensitic. Overall, the mechanical
HIPped at 1203 K (930 C) have significantly better properties that have been investigated here are within
fatigue properties than samples HIPped either at 1153 K the large range obtainable in thermomechanically
(880 C) or 1293 K (1020 C). In these samples, it was processed samples.[11,29]

Table IV. Arithmetic Mean Surface Roughness (Ra) and Maximum Peak to Valley Height (Rmax) of HIPped Four Point Bending
Fatigue Samples with Different Surface Conditions

Conventionally HIPped Electropolished Machined Surface Improved As-HIPped Surface


Ra (lm) 9.08 4.69 0.4 1.05
Rmax (lm) 41.4 28.24 3.75 9.85

1044—VOLUME 41A, APRIL 2010 METALLURGICAL AND MATERIALS TRANSACTIONS A


V. CONCLUSIONS 6. E. Arzt, M.F. Ashby, and K.E. Easterling: Metall. Trans. A, 1983,
vol. 14A, pp. 211–21.
1. The fatigue limit of the as-HIPped surface is signifi- 7. W.X. Yuan, J. Mei, V. Samarov, D. Seliverstov, and X. Wu: J.
cantly improved by using a HIPping schedule, Mater. Process. Technol., 2007, vol. 182, pp. 39–49.
which results in a much smoother as-HIPped 8. F.H. Froes and J. Hebeisen: Hot Isostatic Pressing Int. Conf. 1999,
International Academic Publishing, Beijing, 1999, pp. 1–24.
surface than is obtained with conventional HIP 9. Materials Properties Handbook: Titanium Alloys, R. Boyer, G.
schedules. Diffusion of C from the tooling may con- Welsch, and E.W. Collings, eds., ASM INTERNATIONAL,
tribute to the high fatigue limit of samples contain- Materials Park, OH, 1994, pp. 483–84.
ing the smooth, as-HIPped surface. 10. R.R. Boyer: Mater. Sci. Eng., 1996, vol. A213, pp. 103–14.
11. G. Lütjering and J.C. Willians: Engineering Materials and Pro-
2. A HIPping temperature of 1203 K (930 C) results cesses: Titanium, Springer, Berlin, Germany, 2003.
in the best balance of properties for the conditions 12. I.J. Polmear: Light Alloys: Metallurgy of the Light Metals, 3rd ed.,
investigated here, and these properties are compara- Edward Arnold, London, 1995.
ble with those reported in forged samples. 13. J.C. Williams: Mater. Sci. Eng., 1999, vol. A263, pp. 107–11.
3. The origin of the properties obtained in HIPped 14. A. Davidson, K. Zhang, W. Yuan, J. Mei, M.R. Bache, M.H.
Loretto, W. Voice, and X. Wu: Mater. Sci. Technol., 2006, vol. 22
samples is the martensitic structure within the powder (5), pp. 553–60.
particles, which gives rise to a fine microstructure. 15. British Standard, ‘‘Fracture Mechanics Toughness Tests—Part 1:
Method for Determination of KIc, Critical CTOD, and Critical J
Values of Metallic Materials,’’ BSI BS7448-1, 1991.
16. D.P. DeLo and H.R. Piehler: Acta Mater., 1999, vol. 47 (9),
pp. 2841–52.
ACKNOWLEDGMENTS 17. D.P. DeLo, R.E. Dutton, S.L. Semiatin, and H.R. Piehler: Acta
Mater., 1999, vol. 47 (11), pp. 3159–67.
The authors acknowledge the financial support of 18. J.H. Kim, S.L. Semiatin, and C.S. Lee: Acta Mater., 2003, vol. 51,
this work by ORS and Rolls-Royce plc. Tremendous pp. 5613–26.
thanks are due to Professor M.H. Loretto for many 19. S.L. Semiatin and T.R. Bieler: Acta Mater., 2001, vol. 49,
stimulating discussions and thoughtful suggestions. pp. 3565–73.
20. M.R. Bache: Int. J. Fatigue, 2003, vol. 25, pp. 1079–87.
21. S. Mironov, M. Murzinova, S. Zherebtsov, G.A. Salishchev, and
S.L. Semiatin: Acta Mater., 2009, vol. 57, pp. 2470–81.
22. M.R. Bache: Int. J. Fatigue, 1999, vol. 21, pp. 105–11.
23. C. Kelto: in Powder Metallurgy of Titanium Alloys, F. Froes and
REFERENCES J. Smugeresky, eds., AIME, Warrendale, PA, 1980, pp. 5–9.
1. W.B. Li, M.F. Ashby, and K.E. Easterling: Acta Metall., 1987, 24. D. Novovic: Ph.D. Thesis, The University of Birmingham,
vol. 35 (12), pp. 2831–42. Birmingham, United Kingdom, 2004.
2. D. Seliverstov, V. Samarov, V. Goloveshkin, S. Alexandrov, and 25. C. Arvieu, J.P. Manaud, and J.M. Quenisset: J. Alloys Compd.,
P. Elkstrom: Proc. Conf. on ‘‘Hot Isostatic Pressing,’’ L. Delaey 2003, vol. 368, pp. 116–22.
and H. Tas, eds., Elsevier Science BV., Amsterdam, 1994, 26. X. Wu: Ph.D. Thesis, The University of Birmingham,
pp. 555–60. Birmingham, United Kingdom, 1996.
3. C. Dellis, P.L. Gallo, R. Baccino, and F. Morret: Proc. Int. Conf. 27. Materials Properties Handbook, Titanium Alloys, R. Boyer,
on ‘‘Hot Isostatic Pressing’’, ASM INTERNATIONAL, Materials G. Welsch, and E.W. Collings, eds., ASM INTERNATIONAL,
Park, OH, 1996, pp. 75–79. Materials Park, OH, 1994, p. 64.
4. G. Raisson: Hot Isostatic Pressing Int. Conf. 1999, International 28. J. Mei: University of Birmingham (IRC), Birmingham, United
Academic Publishing, Beijing, 1999, pp. 390–93. Kingdom, unpublished research, 2007.
5. R. Baccino, F. Morret, F. Fellerin, D. Guichard, and G. Raisson: 29. D.B. Lanning, T. Nicholas, and G.K. Haritos: Mech. Mater.,
Mater. Design, 2000, vol. 21, pp. 345–50. 2002, vol. 34, pp. 127–34.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 41A, APRIL 2010—1045

You might also like