Note: Descriptions are shown in the official language in which they were submitted.
CA 02306672 2000-04-06
WO 99/20804 PCT/US97/16812
-
METHOD FOR DISSOLUTION OF
NITROGEN-RICH INCLUSIONS IN TITANIUM
AND TITANIUM ALLOYS
FIELD OF THE INVENTION
5 The present invention relates generally to dissolution of
nitrogen-rich inclusions in titanium alloys. More particularly, the
invention relates to removal of nitrogen-rich inclusions using
electroslag remelting and a halide flux.
to BACKGROUND OF THE INVENTION
Aircraft jet-engine components operating in the
temperature range of about 200 to 500°C are frequently fabricated
from titanium-base alloys. Often, the aircraft components are
subjected to intense levels of cyclic stress and therefore the dominant
15 mode of mechanical failure is low cycle fatigue (LCF). Such failures
start at an initiation site and then grow by crack growth.
tnitiation of the crack is not an intrinsic property of the
titanium alloy. Initiation occurs at an initiation site that is a region that
can be chemically or structurally different from the basic titanium alloy.
2o Initiation sites are characterized as hard or brittle precipitates and
inclusions, voids, combinations of the two, or other different regions.
The larger the initiation site, the longer the initial crack, the faster the
initial crack growth rate and the shorter the low cycle fatigue life. Thus
there is a need to minimize the size of the initiation site.
2 5 Initiation sites for low cycle fatigue failure of titanium-
base alloy jet-engine components frequently occur at nitrogen-rich
inclusions. A nitrogen-rich inclusion is often referred to as a "hard
alpha" inclusion in the titanium-base alloy. The core of the inclusion is
titanium nitride (TIN) which is surrounded by a layer of a-titanium,
3 0 which in turn is surrounded by a layer of b-titanium. In some cases the
core of TiN might be absent and the a-titanium region might be more
extensive.
CA 02306672 2000-04-06
WO 99120804 PCTIUS99/16812 ,
-2-
"Hard alpha" inclusions (nitrogen-rich) are very brittle
when compared to the surrounding titanium-base alloy. As a result,
the areas with nitrogen-rich inclusions are the first to crack under
intense cyclic stress, thus forming an initiation site. To exacerbate the
5 situation further, the presence of the "hard alpha" inclusion frequently
causes a void to form during forging or hot forming, thus increasing
the size of the potential initiation site still further.
The overall process for titanium-base alloys, starting with
the ore and ending with the component assembled into an engine,
1o takes many steps. The first step is the enrichment of the titanium
dioxide from ores, followed by production of titanium tetrachloride, the
reduction of titanium tetrachloride to titanium sponge, the preparation
of the primary titanium-base alloy electrode, followed by multiple arc
meltings or arc plus hearth melting, and fast thermomechanical
15 processing of the ingot.
The nitrogen-rich inclusions of concern in the titanium-
base alloys are created as defects during the refining steps of the
process and are not successfully dissolved during the melting steps,
such as the preparation of the primary electrode, the multiple arc
2 o meltings or arc plus hearth melting, and the thermomechanical
processing of the ingot. Chopping the refined titanium into smaller
pieces and using melting schemes with longer melt residence times,
such as triple arc melt and hearth melt, have helped, but can not
guarantee elimination of inclusions. Currently, ultrasonic inspection is
25 being used at several stages along the processing route after melting
to find and discard materials containing nitrogen-rich inclusions. This
is time consuming and expensive.
For the above reasons, there is an increased need to
eliminate or minimize the size of nitrogen-rich inclusions in titanium
3 o and titanium-base alloys.
SUMMARY OF THE INVENTION
This invention is directed toward a process for the
electroslag remelting of titanium and titanium alloys which will ensure
CA 02306672 2000-04-06
WO 99/20804 PCTIUS97/16812
-3-
that nitrogen-rich inclusions are dissolved or at least minimized during
the melting process. The approach is a derivative of electroslag
remelting (ESR) applied to titanium-base alloys.
Briefly, electroslag remelting can be described as a
5 process where the material to be refined, the electrode, is melted by
passing a current through it into a molten flux or slag, which is
resistively heated and which, in turn, melts the electrode. Molten
metal forms on the end of the electrode and falls as droplets through
the flux, forming an ingot in a cooled crucible. The process continues
l0 until the electrode is consumed and the refined metal ingot is formed.
In this invention, the halide flux, also referred to as slag,
is primarily a calcium halide, such as calcium fluoride (CaF2) or
calcium chloride. Calcium fluoride is preferred. The halide flux, such
as calcium fluoride, has suitable additions, such as dissolved metals
15 and oxides. Calcium metal is a preferred addition and is sufficient in
the slag in a sufficient amount to aid in lowering the partial pressure of
nitrogen and oxygen in the flux, while increasing the conduction of the
flux.
The term "nitrogen-rich inclusionu means a titanium
2o nitride (TiN) core surrounded successively by a layer of alpha-titanium
(a-Ti) and a layer of beta-titanium (b-Ti), with decreasing nitrogen
concentration and decreasing chemical activity of nitrogen from the
center of the inclusion to its outer surface. The nitrogen-rich inclusion
is surrounded by the b-Ti electrode, which typically has a nitrogen
2 5 content of approximately 50 wppm.
Dissolution of the nitrogen-rich inclusion takes place by
transport of titanium and nitrogen from the electrode into the flux as
the flux flows past the exposed surface of the titanium nitride inclusion.
The exposed surface of the titanium inclusion is located on the face of
3 o the electrode that is in contact with the molten flux. The nitrogen is
transported to the liquid metal film on the face of the electrode, where
it is re-absorbed. Because nitrogen is continually re-absorbed by the
liquid metal film, the chemical activity of nitrogen (represented by its
partial pressure) in the flux is maintained lower than that in the
CA 02306672 2000-04-06
WO 99/20804 PCT/US97/16812
-4-
inclusion, but higher than that at the liquid metal film/flux interface. As
a result, thermodynamic driving forces in the liquid metal film and in
the flux always exist to transport nitrogen away from the inclusion.
The nitrogen partial pressures in equilibrium with different regions of
5 the nitrogen-rich inclusion and the liquid metal film on the electrode
face can be calculated.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a schematic view of the electrosfag refining of a
1o titanium alloy electrode showing the nitrogen-rich inclusion at the
electrode/flux interface.
FIG. 2 is a schematic view of dissolution of the nitrogen-
rich inclusion in the flux based on the partial pressures of nitrogen.
15 DETAILED DESCRIPTION OF THE INVENTION
It has been discovered that electroslag remelting of
titanium metal or alloys under controlled partial pressures of nitrogen,
and somewhat oxygen, can dissolve nitrogen-rich inclusions to help
eliminate or reduce initiation sites for cracks.
2 o One aspect of the invention is a method of removing
nitrogen rich-inclusions from a titanium containing electrode,
comprising the steps of: contacting a bottom surface of the titanium
containing electrode with a flux in a crucible; passing a sufficient
amount of electric current through the electrode and flux to melt the
25 bottom surface of the electrode while resistively heating the flux at a
temperature that melts the bottom of the electrode; and dissolving the
nitrogen-rich inclusions exposed to the flux by maintaining a nitrogen
partial pressure in the flux lower than that in the inclusion.
Another aspect of the invention is the article made by the
3 o above method.
Still another aspect of the invention is a method to refine
titanium or titanium alloys by electrosiag refining, comprising the steps
of: heating in a non-oxidizing atmosphere a calcium halide flux
containing about 1 - 4.5 weight percent calcium metal to a temperature
35 above about 1600 C; stirring the flux with a magnetic or arc stirring
CA 02306672 2000-04-06
WO 99120804 PG"TIUS97/16812 .
-5-
means; touching the flux with a titanium or titanium alloy electrode
containing nitrogen-rich inclusions; applying an alternating current to
the titanium or titanium alloy electrode, which passes through the flux
to resistively heat the flux; and maintaining a partial pressure of
5 nitrogen in the flux below the partial pressure of nitrogen in the
inclusion.
The calcium metal in the flux helps to provide low partial
pressures of nitrogen and oxygen while increasing the conduction of
the flux. A preferred range for the partial pressure of nitrogen in the
1o flux is about 10''2 to 10''5 atmospheres and for oxygen about 10'2°
to
10'25 atmospheres. Decreasing the flux resistance aids the elevation
of the temperature of the flux, which in turn provides the appearance
of the generation of microarcs on the surface between the flux and the
end of the consumable electrode. Nonmetallic inclusions are crushed
15 by these microarcs, partially or fully.
The low partial pressure of nitrogen changes the
stoichiometry of titanium nitrides. For example, titanium nitride, TiN,
may be transferred to TiN°.se or TiN°.ss. The low partial
pressure of
oxygen increases the sorption characteristics of the nitrogen in the
2 o flux.
Illustrated schematically in Figure 1 is a view of the
electroslag remelting system depicting titanium or titanium alloys with
a nitrogen-rich inclusion. The system 10 includes a titanium or
titanium alloy electrode 12 with a nitrogen-rich inclusion 14 at the
25 interface 16 between the electrode 12 and the halide flux 18. The
electrode 12 forms molten metal droplets that fall through the flux 18
and are collected as liquid metal 20. The system includes a crucible
22 in which is suspended the electrode 12, the flux 18, and liquid
metal 20. Conventional means are provided for melting the bottom
3 o end 24 of the electrode 12 as it is fed into the crucible 22. The
heating means include a suitable alternating current power supply
electrically joined to the electrode. Electrical current is carried through
the electrode 12 and through the flux 18, in liquid form, to the crucible
CA 02306672 2000-04-06
WO 99120804 PCT/US97/16812 .
-6-
22. In this way, the flux 18 is resistively heated to a suitably high
temperature to melt the bottom end 24 of the electrode.
The electrode 12 or otherwise known as the ingot to be
refined or remelted, is formed of titanium or any suitable titanium alloy
5 requiring electroslag remelting. Examples of titanium alloys are Ti
6Al-4V, Ti-6AI-2Sn-4Zr-2Mo, and Ti-17-(5AI-4Mo-4Cr-2-Sn-2Zr). A
suitable flux 18 is a halide flux, in particular calcium fluoride. The flux
may have additional suitable additions, such as dissolved metals and
oxides, including calcium metal in a preferred amount of about 1 to 4.5
1o weight percent.
Now referring to Figure 2, the nitrogen-rich inclusion 14
is composed of an inner core of titanium nitride (TiN) 26, followed by
alpha-titanium section 28, and an outer section called beta-titanium
30. The nitrogen-rich inclusion 14 can be up to about 1200
15 micrometers in diameter, and often is in the range of about 300 to
1000 micrometers in diameter. The end 24 of the electrode 12 is a
liquid titanium film about 30 to 100 micrometers thick. The titanium
electrode 12 generally contains up to about 50 weight parts per million
(wppm) nitrogen. This is not the nitrogen-rich inclusion 14.
2o As the nitrogen-rich inclusion 14 moves through the
liquid film 24 on the electrode i2, the beta -titanium 30 on its front face
equilibrates with the liquid film 24. For instance, if the superheat in
the liquid film 24 is 10 °C, the beta-titanium 30 at the inclusion/flux
interface 32 has an effective nitrogen partial pressure corresponding
25 to saturated liquid in equilibrium with beta-titanium at that temperature.
The rest of the liquid film 24 is undersaturated at 50 wppm nitrogen
and corresponds to a lower nitrogen partial pressure. The
decomposition nitrogen partial pressure for stoichiometric titanium
nitride (TiN) is approximately 10-'° atmospheres. The driving force for
3 o nitrogen transfer from the inclusion into the flux is the two orders of
magnitude difference between the nitrogen partial pressure at the
inclusionlflux interface 32 and the nitrogen partial pressure at the
liquid metal film/flux interface 16. The nitrogen partial pressure in the
flux 18 must fall between these two values. tf the driving force shown
CA 02306672 2000-04-06
WO 99120804 PG"T/US97116812 _
-7-
is not adequate, the inclusion will protrude from the liquid metal
surface, and its temperature will increase. The nitrogen partial
pressure corresponding to liquid/beta-titanium equilibrium will increase
with increasing temperature until a sufficient driving force is available
5 to remove nitrogen from the inclusion at an adequate rate. As the
lower-nitrogen outer regions of the inclusion are removed, the nitrogen
chemical activity or partial pressure in 'the exposed inclusion
increases, thus the nitrogen removal rate should increase as the
inclusion continues to dissolve.
10 It has further been discovered that several factors must
exist for the nitrogen transport process to be effective. First, an
adequate thermodynamic driving force must exist near the beginning
of the process, when the concentration difference between the outer
layer of the inclusion and the liquid metal film on the electrode is
15 small. Second, the chemical activity of nitrogen in the flux must fall
between that at the inclusion/flux and liquid metal filmlflux interfaces.
Third, the nitrogen solubility in the flux and the flux velocity must be
adequate so that the flux does not come into thermodynamic
equilibrium locally with the inclusion, and so that nitrogen is
2 o transported at an adequate rate from the inclusions to the liquid metal
film on the electrode. Fourth, the nitrogen removal rate from the flux
into the liquid metal film by interfacial transfer and diffusion in the
liquid metal must be high enough to keep the process going. Fifth, the
nitrogen capacity of the liquid metal layer must be high enough to
2 5 absorb all of the nitrogen transported from the inclusions.
The nitrogen may go into solution as an ion, as a neutral
dissolved species, or as a complex ion with oxygen or some other
species. Rapid circulation of the flux to move nitrogen away from the
inclusion is suggested. The flow rate of the flux past the inclusion is
3 o important and can be augmented independently of temperature by arc
and magnetic stirring.
It is essential that the various kinetic processes listed
above are fast enough so that the nitrogen-rich inclusion dissolves at
a rate at least as fast as the melting back of the electrode face. Since
CA 02306672 2000-04-06
WO 99120804 PCT/US97I16812
-g-
the chemical activity of nitrogen in the flux is low, it is also essential
that additional nitrogen is not introduced into the flux from the
surrounding atmosphere. An inert gas, which may have to be Bettered
to produce the required low nitrogen partial pressure, must be
5 maintained above the exposed flux. Generally, the electrode melting
is done in a non-oxidizing environment, such as argon, helium, neon,
hydrogen or mixtures thereof.
Additions to the calcium halide-based flux which would
aid in nitrogen dissolution may also be beneficial. Oxygen in the form
10 of a dissolved oxide such as calcium oxide (Ca0) or titanium oxide
(Ti0), might be used to oxygenate the center of the inclusion, reducing
its melting temperature and enhancing diffusion and dissolution.
Solubility of calcium and oxygen in calcium-rich calciurr~ halide, such
as calcium chloride (CaCl2) fluxes has been demonstrated.
15 Thermodynamic calculations have been used to indicate
that dissolution of a nitrogen rich-inclusion is possible by electroslag
refining (ESR}. The results of these calculations are shown
schematically in Figure 2. The titanium nitride core of tile inclusion is
surrounded successively by a layer of alpha-titanium and a layer of
2 o beta-titanium, each layer with a decreasing nitrogen concentration and
decreasing chemical activity of nitrogen. The inclusion is shown
embedded in a beta-titanium electrode, which typically has a nitrogen
content of approximately 50 parts per million by weight (wppm). It is
demonstrated that dissolution of the inclusion will take place by
25 transport of nitrogen into the flux, as the flux flows past the exposed
surface of the inclusion. The nitrogen will then be transported to the
liquid metal film on the face of the electrode, where it will be re-
absorbed in a widely dispersed fashion. This layer will melt, and
droplets will pass through the flux to form a solidified ingot below. For
3 o this approach, the electroslag refining system must be capable of
maintaining a partial pressure of nitrogen in the flux and in the
atmosphere above the flux of less than 10''5 atmospheres.
CA 02306672 2000-04-06
WO 99120804 PCT/US97/16812
_g_
Examples:
Example 1. The interaction of titanium nitride (TiN) with a calcium
flouride-base flux has been observed in the laboratory, by immersion
of a sample of hot isostatic pressed titanium nitride (TiN) in a 70%
5 calcium fluoride (CaF2), 15% calcium oxide (Ca0), 15% aluminum
oxide (AI203) flux. The flux was heated in a graphite crucibie by
induction heating. Breakup of the titanium nitride ( TiN) was noted.
Example 2: tnclusion Preparation - Two types of inclusions were
1o prepared by GE-CRD. Inclusions with Identification Number 30 were
fabricated by placing nitrided sponge and titanium powder in titanium
alloy tubes. The tubes were sealed with plugs by electron beam
welding. The tubes were then isostatically pressed at 1200~C I 1000
atm. / 4 hr. Samples were cut from these tubes with 12.5 mm diameter
15 by 10 mm length. The nitrogen containing portion of each sample was
10 mm diameter.
Inclusions with Identification Number 49 were fabricated
by placing titanium nitride powder in titanium alloy tubes. The tubes
were sealed with plugs by electron beam welding. The tubes were
2 o then isostatically pressed at 1200°C I 1000 atm. I 4 hr. Samples
were
cut from these tubes with 12.5 mm diameter by 10 mm length. The
nitrogen containing portion of each sample was 10 mm diameter.
Inclusion information is summarized in Section 1 of the
attached table 1.
25 Electrodes were prepared by pressing titanium sponge
into cylinders 60 mm in diameter by approximately 250 to 275 mm
long. Two such cylinders (plus a top attachment stub) were welded
together by TIG welding to form an electrode. Holes (13 mm diameter)
were drilled into the sides of the electrodes to a depth of
3 o approximately 13 mm. Inclusions were inserted into these holes and
welded into place by TIG welding. The inclusion types and positions
(measured from the bottom of the electrodes) are summarized in
Section 2 of the attached table 1.
CA 02306672 2000-04-06
WO 99/20804 PCT/US97/16812 .
-10-
The electroslag refining system used for these examples
was of the type having a water cooled crucible, approximately 95 mm
inside diameter by 360 mm long and capable of producing an ingot up
to 200 mm in length. The electrode was placed inside a water cooled
5 chamber mounted above the water cooled crucible. The crucible and
chamber were evacuated and then back filled with argon before
melting. The system was kept at a positive pressure of argon of 0.15
atmospheres during melting. A small flow of argon occurred to
compensate for leakage out the various seals in the system. The flux
1 o used for these experiments was made up from 1000 grams of calcium
fluoride and 20 grams of metallic calcium. Calcium vapor rises into
the chamber above the crucible and reacts with the small amounts of
nitrogen (and oxygen) in the argon to reduce the partial pressure of
nitrogen in the atmosphere above the flux to less than the required 10'
15 'S atmospheres.
A typical melting run consisted of a starting period (at
low current) to melt the flux, followed by a melting period (at higher
currents) to melt the electrode. Melting was carried out at the melting
currents summarized in Section 3 of the attached table 1. A time of
2 o about 7 to 10 minutes was required for the completion of melting. An
average grain size of about 2 to 4 mm was observed on the top
surface of these ingots. The refined ingots are the metal that has
been collected after electros lag remelting of the electrode.
The ingots were machined on the circumference and the
25 top and bottom surfaces to prepare them for ultrasonic inspection. A
Model 11 YA Ultrasonic Defect Tester was used. This instrument was
capable of frequencies of 1.25, 2.5, 5 and 10 MHz. A frequency of 2.5
MHz was selected for this examination. Inspection was conducted
with the transducer placed on either the top or the bottom surface of
3 o the ingot. A drop of glycerin was placed between the transducer and
the ingot surface to provide good contact. In order to adjust the
instrument, a calibration block was prepared from a pure titanium
ingot, with 1.5 and 2.5 mm diameter holes drilled into the side at three
locations along the length of the ingot. The instrument was calibrated
CA 02306672 2000-04-06
WO 99/20804 PCTNS97/16812 .
-11 -
to give approximately the same amplitude of indication from each of
the three holes, whether viewed from the top or viewed from the
bottom of the ingot. These adjustments then remained unchanged
during measurement of the five ingots prepared from electrodes
5 containing inclusions.
All ingots showed some small indications in the top 20
mm of length, particularly at the centerline. It was assumed that these
indications were from shrinkage porosity formed by rapid solidification
at the top of the ingot, when the melting current was turned off.
to At least one major ultrasonic indication was observed in
each ingot. The number of indications and positions (measured from
the bottom of the ingots) are summarized in Section 3 of the attached
table 1. All indications, except for one, occurred at positions in the
lower 15 to 30% of the length of the ingot. Their position can not be
15 correlated with the positions of the inclusions in the electrodes. It is
assumed that they represent disturbances caused when the melting
current was changed from the low current of the starting period (flux
melting) to the high current of the melting period (electrode melting).
In ingot D6, one large indication occurred (at 58 mm from the bottom)
2 o near the location of the lower inclusion in the electrode that was
melted. This particular inclusion was only welded to the electrode in
two places and may well have fallen into the liquid metal pool during
melting.
The results from the preliminary ultrasonic inspection
2 5 showed that there are no large inclusions in the ingots.
CA 02306672 2000-04-06
WO 99/20804 PCT/US97/16812
- 12 -
0 00
M
st O M N
M ~
~ ~
E ~f71.n~ ~ M N
N O
E N N N tn a
O
C
Z Z '_p e-.,..e-M d'
41~ ~e-i-'~
Zi
O O
v s
U U 0 MM M M
.
E c~o~rio o ~
- ~ ~ ~
o c ,
a ~N N N
!~
G
O
M'Od'M ~
a ~ ~..~ oo oio ~ c~a~~nco~
Q i~ N MM thM Z Z Z Z Z
Z - ~
~z
~ ~ r UM M O "E M N M N ~
E ~nL~n~ ~ ~ .-~ ~ .-~
~n E
o ~ ~ -~
'o >
v u~ ~
E ~ co ~ o
~ ~E o 0 0 0 0
0
a ~ '
'
o d ~
~ o o !3 ~
E M ~ C
w
i ~ z zz z z ~ ~? N C~v ~ co
~
Z 2
N lil r~
SUBSTITUTE SHEET (RULE 26)
