PHASE TRANSFORMATIONS IN BETA ISOMORPHOUS TITANIUM ALLOYS*

M. K. KOULt$ and J. F. BREEDIS?

Formation of the omega and alpha phases on quenching and aping and of martensite under an applied
stress has been studied in binary, beta isomorphous titanium alloys containing molybdenum, niobium
and vanadium. Thermodynamic catculations using the regular solution approximation reveal that the
beta phase exhibits a tendency for phase separation. Such compositional fluctuations in the beta phase
can lead to an increase in the driving force for alpha phase formation in titanium-vanadium but arc
not associated with omega formation. Stress-induced martensite formed in molybdenum- and vanadium-
containing alloys is found to be hexagonal close-packed, as expected from thermodynamic considerations.

TRA~SFOR~l_4TIO~S DE PHASES DAXS LES ALLIAGES DE TITASE ISO~IORPHES BETA

Les auteurs ont etudie la formation des phases omega et alpha par trempe et vieillissrment, et la
formation de martensite par application d’une contrainte dans ler alliages de titane binaires isomorphcs
beta contenant du molybd&ne, du niobium et du vanadium. Lsa calculs thermodynamiques utiliaaut
l’approximation d’une solution reguliere revelent clue la phase beta presentr une tendance 1 la separation
des phases. De telles fluctuations de composition dans la phase bet,a peuvent conduire h un accroissement
tie 18 force directrice pour la formation de la phase alpha dans le titan-vanadium, ma& ne sent par
8ssociQes k la formation de la phase omega. Les auteurs ont trouv6 que la martensite induitr par appli-
cation d’une contrainte et form&e dans les alliages eontenant du mo1pbdPn.eet du vanadium est hesaponalr
comparte, comme l’avaient prcivu les calculs thermodynemiques.

PHASESUMWASDLUSGES IS BETA-ISOMORPHES TITAS-LEGIERUSGES

Die Bilclung von omega- und alpha-Phasen beim Abschrecken und -4ltern und die Bildung von Martensit
unter Last. wurde in bin&en, beta-isomorphen Titan-Legierungen mit Molybdiin-, Siob- und Vanad-
iumzusiitzen untersucht . Thermodynamische Berechnungen in der Eaherung der repuliiren L&ung
, da@ die beta-phase eine Tendenz zur Pha~ntreMung aufweist. Sol&e Fluktuationen der
~~~menset.~un~ in der beta-phase kiinnen zu einer Zunahme der treibenden &aft fiir die Bildunp der
alpha-Ph8se in Titan-Vanadium fiihren, hangen aber nicht mit der Bildung der omega-phase zusam-
men. Der auf Grund von Spannungen in Legierungen mit Molybdan- und Vanadiumanteilen gebildete
Martensit ist, wie man es aus thermodynamischen Uberlegengen erwartet, hexagonal dirhtcst gepaokt.

Introduction phase is retained on cooling to room temperature in

The phase boundaries of beta isomorphous titanium region III and will not transform to omega during
alloy systems have not been determined experimen- aging. Further, it is usually believed that t,he beta
tally below 600°C because of the difficulty in attaining phase remains stable with increasing solute concen-
equilibrium. A more complete understanding of such tration and that no fuxther transformabions occur in
binary titanium alloys containing molybdenum, these alloy systems.
niobium or vanadium is ofimportance because of their
EXPERIMENTAL METHODS
technological applicat’ions.
The section of the beta isomorphous phase diagram
shown in Fig. 1 can be subdivided into three regions The alloys studied, their designation and analyzed
according to the type of stable and n~etast.ablephases conl~sitions are given in Table 1. Unalloyed titan-
which can occur. The high-temperature. b.e.c. bet.a ium, and titanium-26 wt.7 o molybdenum rods were
phase can transform to hexagonal close-packed, produced by double consumable-electrode arc-melting,
equilibrium alpha phase (51) and martensit,e (a’) in followed by hot forging, hot, and cold swaging, and
region I. In region II. beta can be completely re- centerless grinding. The remaining compositions
tained at‘ room temperature in a met&&able state b> were prepared by non-colls~~~~lable arc-melting. High-
rapid quenching or it can partially decompose to purit,y, electrolytic-grade t,itanium sponge, molyb-
another structure, the omega phase. during quenching denum (99.99,6). niobium (99.V;) and vanadium
or isothermal annealing at temperatures below 400%. (99.87;) were used. Fifteen gram buttons of each
In addition, stress-induced martensite (Y-“) can form alloy were melted at least twice under an argon atmos-
under stress in certain alloys in t,his region. The beta phere. All alloys were liomoge~lize(l at 1100°C
* Received September 8, 1363: revised Sovrmbcr 5, 1303. followed by quenching and cold rolling to a thickness
This paper is abstracted from a dissertation by 11. K. Iioul of 0.015 in.
submitted to the Department, of Metallurgy and Materials
Science, in partial fulfillment, of the rcyturc~ments for the ,411 heat treatments were performed in quartz
degree of Doctor of Philosophy (136H). capsules evacuated to a pressure of about 10-5mm
t Center for Matcriais Science and Engineering, Massachus-
etts Institute of Technology. Cambridge. ~lassachusetts of mercury, repeatedly purged w&h ultra-high purity
32139, U.S.A. argon and refilled with argon to a pressure of nearly
$ Now at: Research and Development Department, Union
Carbide Corporation, Siagara Falls, Sew York, 14302, l?.S.A. 4 atm at room temperature. Specimens were quenched
ACTA JIETALLURGIC.4, VOL. 18, Jr-SE 1970 579
&?o ACTA SIETALLURGICA, VOL. 18, 1970

with t,he work in titanium-vanadium by Silcock and

1 ; II / m coworkers”g2) where the omega phase is best reprr-
I /
I
sented by an hexagonal unit cell having an axial ratio
of 0.61. Each cell contains 3 atoms, at positions (000)
and (I-, 8, X) where X = 0.50. Additional confirma-
tion of the hexagonal nature of the omega phase has
been made by Blackburnand T~7illiams(3) and Brammer
and Rhodes@) in titanium-molybdenum and titanium-
niobium alloys, respectively. Further. all diffraction
patterns obtained are compatible with the orientation
relationship: (ill), /I (OOOl), and [liO], /I [llzo],.
A selected-area diffraction pat,tern obtained from
quenched titanium-16 vanadium is shown in Fig. 2’.
):
The electron beam is parallel to the [Sill, direction
II in this pattern which cont.ains reflections from two
/
I
‘ !
variants of the omega phase. Tilt.ing of the foil about
11
II the [121], direction markedly decreases the intensities
I
Ti of the circled reflections which appear to arise from
PERCENT SOLUTE multiple scattering. However. the intensities of the
Ffa. 1. Beta isomorphous phase diagram for titanium
alloys showing three regions which undergo characteristic 1010 and 2020 reflections, forbidden by the structure
trttnsformation behavior. factor in aged alloys according to Rilcock.‘z) were
after heat treatment by breaking the capsules in iced not markedly affected by tilt>ing up to 20” from the
water. foregoing orientation (Fig. 4a). A reflection such as
1OiOcan be derived by multiple IOil,-type refle&ions.
Electron microscopy
however. Such multiple scat,tering is possible where
Foils for transmission electron microscopy were the electron beam encount,ers many small particles in
prepared by electropolishing with two electrolytes; passing through the foil. Furthermore, beta phasr
methanol (150 ml), n-butyl alcohol (90 ml) and per- reflections remain circular while omega phase refiec-
chloric acid (15 ml) was used for all alloys except tions are elongated along respective (lOiO>, directions
titanium-niobium for which an electrolyte composed in each variant. The sense of reflection elongation is
of methanol (280 ml), sulfuric acid (15 ml) and hydro-
found not to vary by tilt,ing t.ox-ards a [%O], orienta-
fluoric acid (7.5 ml) was employed. All electropol- tion (Fig. 4a). Thus, reflections sre elongated in the
ishing was performed below -40°C to prevent hydride [OOOI], of each variant.
formation. Dark-field microscopy was accomplished
The appearance of beta alloys which contain a
by deflection of the incident beam to produce aberra-
large volume fraction of omega aft’er quenching is
tion-free images.
illustrated by Fig. 3. Small omega phase particles do
EXPERIMENTAL RESULTS
not, exhibit a distinct morphology in quenched alloys.
The omega phase Particles are clearly evident. only; near extincti~~n
Formation of omega during quenching and aging, contours indicat,ing that both struct,nrc Pa&or and
as well as its structure and morphology. were studied strain contrast can operate. providt~d the omega phasr
using selected-area electron diffraction ancl dark-field is initially coherent,. Verification of the cbsistencr of the
microscopic techniques in several titanium-vanadium latter contribution is complicatrd by the larpo volumcb
alloys. Indexing of diffraction patsterns is compatible fraction of particles.

UnRllO\-e~l
tit,allium - - - 370 40 x0
Ti-13 XIo 12.9 - - - -
Ti-26 310 26.4 - - 870 - -
Ti-65 Sb - 65.0 - - - -
Ti-16 \- - - 16.2 850 190 101,
Tl-50 V - - 49.5 800 - -
 KOUL AWD BREEDIS: PHASE TRANSFORMATIONS IS TITANIUM ALLOYS 581

FIG. 2. Selected-area electron diffraction pattern ob-

tained from quenched titanium-16 vanadium. Electron
beam is parallel to @ii]@ and (2n31w of two variants of
the omega phase.

It is concluded from the present studies of Alseries

of titanium-vanadium alloys containing between 15
and 50 wt.:& vanadium that omega forms on quench-
ing in alloys containing up to about 20% vanadium
and only after aging at temperatures below 400°C
in alloys containing between 20 and 3Oo/o vanadium.
Alloys of higher vanadium contents do not transform
to the omega phase. It is difficult to quantitatively
estimate the amount of the omega from thin foil ob-
servations because of particle overlap. However, it
is clear that the total volume fraction of the omega
phase which forms on aging decreases with increas-
ing vanadium content.
Aging at elevated temperat,ures below 4OWT leads
FIG. 4. Selected-area electron diffraction patterns ob-
tained from titanium-16 vanadium: (a) after quenching,
and (b) aft,er aging for 260 hr at 350°C. Electron bram is
parallel t.o [2lO]p in both cases.

to the elimination of multiply-diffracted omega

reflections attending coarsening of these precipitates
and t,he eventual formation of the equilibrium alpha
phase (Fig. 4). The lOi0, reflections are aligned along
<112), directions, which is compatible with the orien-
tation relationship of parallelism between close-packed
planes and directions of these phases. Omega particles
develop a characteristic cuboidal morphology during
aging as shown in Figure 5(a). After aging for 260 hr
at S~O”C, omega particles have attained their maxi-
mum size of around 1500 A measured along an edge,
PIG. 3. Fine dispersion of omega particles in titanium-16
vanadium quenched from 1000°C. !Ihe bands lying between adjacent omega particles
589 ACTA METALLURCICA, VOL. 18. 1970

(Fig. 5b) contain the beta and equilibrium alpha

phases. The latter are illustrated by the dark-field The possible occurrence of phase separation was
micrograph (Fig. 5~) obtained using the circled examined in beta phase titanium alloys aged for
lOTO, reflection shown in Fig. 4(b). Alpha develops prolonged periods within the reported beta phase
as ellipsoidal precipitates. The absence of strain region. Aging titanium-50 (48.5 at.%) vanadium at
contrast shows that the various phases are not coher- 500% resulted however in rod-shaped, alpha particles
ent after aging. within beta grains (Fig. 6a). A different morphology
of alpha particles is found to occur during aging at
350°C, which is within the published boundaries’b’ of
the alpha-beta region extrapolated from 600°C
(Fig. 6b). The alpha precipitates formed at this lower
~m~rature appear as thin discs lying on (lOO)P planes.
No precipitation was observed on the other hand
within beta grains of titanium-26 molybdenum aged
for prolonged periods at 350% (Fig. 7). However,
this alloy is within the reported two-phase region of
this system after extrapolation of phase boundaries,
and some poorly-defined structure, possibly the alpha
phase, appears at grain boundaries. In contrast to
this observation, a large volume fraction of the alpha
phase was reported(3) in titanium-l 1.5 molybdenum
which also contained omega, during aging at this
temperature for a significantly shorter period of time
(144 hr). Alpha precipitates were also observed to

FIG. 5. Aged titanium-16 vanadium (260 hr at 350%)

showing: (a) cub&da1 omega phase particles viewed iu
dark-field using a IOfi, reflection, (b) Bheabove region
viewed in bright-field. (c) alpha particles viewed in dark- FIG. 6. Aged titanium-50 vanadium showing alpha pre-
field using a lOTO= reflection. cipit.ates after: (a) 40 days at 5OO”C, (b) 30 days at 350%.
 KOUL AND BREEDIS: PHASE TRANSFORMATIOSS IS TITANIUM ALLOYS 583

diffraction rings from a f.c.c. structure (lattice param-

eter of 3.65 A) was observed. This structure does
not correspond to the f.c.c. hydride reported(6,7) for
titanium-niobium alloys, or with any other reported
hydride for that matter. However, no alpha precipita-
tion was observed in this niobium-bearing alloy aft,er
prolonged aging.

Stress-induced marten&e
Stress-induced transformation was studied in meta-
stable titanium-13 molybdenum and tit#anium-16
vanadium. In both cases, stress-induced martensite is
found to be hexagonal close-packed in structure and
not a body-centered structure as previouslyreported.(3)
FIG. 7. Titanium-26 molybdenum aged at 350°C for 30
days. This is illustrated by Fig. 9 and the accompanying

nucleate at grain boundaries in the latter alloy.

Clearly, the beta phase is more rapidly stabilized by
the addition of molybdenum over vanadium.
Finally, titanium-niobium alloys containing be-
tween 50 and 80% niobium were aged at 450°C
for prolonged periods. The published phase diagram
is not known with certainty beyond 25% nio-
bium. Should phase separation be possible, the small
elastic misfit strain between solute-rich and solute-poor
regions would result in little suppression of separation
by spinodal decomposition. The structure observed
in titanium-65 (50 at.%) niobium is shown in Fig. 8.
The selected-area diffraction pattern (insert) shows
only reflections from the beta phase of this alloy.
While the dark patches may represent niobium-rich
beta, it is also possible that these are merely artifacts
produced during electropolishing. Samples prepared
with a related hydrofluoric-sulfuric-lactic acid elec-
trolyte also exhibited a similar structure. Occasionally,

FIG. 9. Stress-induced martensite in: (a) titanium-13

molybdenum, and (b) titanium-16 vanadium. The
electron beam is parallel to [2423]a* in both cases.

selected-area diffraction patterns (insert). Further,

the internal structure of martensite plates eonsists
of a high-density of randomly arranged dislocations
rather than the stacking faults, twins and moire
contrast observed in spontaneously transformed
martensite.
The growth of omega on aging was investigated to
further confirm that this structure was stress-induced
FIG. 8. Titanium-65 niobium aged at 450°C for 30 days. martensite and not merely deformation twins in the
is!34 ACTA METALLURGICA, VOL. 18, 1970

energies of the alpha and beta phases are given by:

G,(z) = (1 - r)G,*’ $ zG,~ + G,z + RT

x [zlns + (1 - z)ln (1 -s)] (la)
Gs(x) = {l - x)GgTi -j- zGpM + G,z + RT
x [zlns -/- (1 - x) ln (1 - x)] (lb)
where GdTi = free energy of pure hexagonal cloee-
packed titanium,
GBTi= free energy of pure b.c.c. titanium,
GBM= free energy of pure b.c.c. solut,e, and
GZBf= free energy of pure hexagonal close-
packed solute.
FIG. 10. Titanium-16 vanadium containing omega pre- The hexagonal close-packed structure is not stable in
cipitates after aging at 350°C for 300 hr. Precipitates in molybdenum, niobium or vanadium. The excess free
contrast using a lOT1, reflection.
energies of mixing G,” and GBE in the system Ti-M
are given in terms of the interaction parameters,
beta, phase. Growth of the omega particles should
-4 and B, by As (1 - 5) and Bx (1 - x) for the alpha
differ in hexagonal close-packed martensite and
and beta phases, respectively. At equilibrium in the
the beta phase. However, no difference in precipitate
two-phase (a + /l) region of the phase diagram.
dimensions is expected between regions of a twinned
structure. Titanium-16 vanadium was aged at 350°C
for about 399 hr after quenching to room temperature
and straining to about 10 per cent. The difference
in the growth of the omega phase is clearly seen in G&$ - GJx.) = (x8 - x,) 2 (3)
C7
Fig. 10. Well-developed ouboids are evident in the
dark-field micrograph of the bets, phase region, whereas where x# and xg are the atomic fractions of solute at
there is very little growth of omega in the adjacent the limits of solubility at a given temperature. From
maitensite plates. these equations, the following expression is obtained:

DISCUSSION
~~~-4- xB2B = RT In s + (GBTi- G,=) (4)
ThermodylLamic considerations (t
The equilibrium phase boundaries, and the phase Two problems are associated with using this expression
transformations which can occur in beta isomorphous in calculating the interaction parameters. First,
systems can be rationalized using the following ther- the values of x= and xP are not known very accurately
modynamic approach. In particular, some insight can for these beta isomorphoua systems. Second, the
be gained into tr&n~orrn&t~onswhich can be expected coefficient of A in equation (a), or xSz, is very much
at’ temperatures below 600% where experimental data smaller than the corresponding coefficient of B.
are lacking. This can lead to large error in solving for these inter-
The heats of mixing for the titanium-molybdenum, action parameters. Neglecting x#z in equation (4)
titanium-niobium and the titanium-vanadium sys- and substituting values for x, and xP at one tempera-
tems can be obtained using t,he regular solution ap- ture in the resulting equation, the sign and magnitude
proximetion and the known portions of the phase of the interaction parameter can be estimated. The
diagrams. Krisement(8) has previously made such a results of such a c&l~ul~tion at 973”K, using the
calculation for the titanium-v~n&dium system. It previously calculated(s) value of (GgTi - G.*‘). or
should be noted that in his calculation to obtain an 177 cal/g-atom, are summarized in Table 2. The
interaction parameter for the beta phase, he assumed TABLE 2. Interaction parameters (B) of the beta phase at
that, [(GBTi- GaTi) - (Gprv- G,“)] is negligible. 973°K
On this basis, he predicted a miscibility gap whose Alloy
critical temperature is 850°C. This assumption is not system
valid since AGF_, is nearly zero while the AC;_= term Ti-MO 0.003 0.11 -211 13109
is theoretically about 2000 caljg-atom. In an alloy Ti-Xb 0.015 0.20 -405 16700
Ti-V 0.015 0.14 -266 +4500
Ti-M (where M represents the solute metal) the free
 ROCL ASD BREEDIS: PHASE TRASSFORMATIOSS IS TITASIVM ALLOTS .5x5

results show positive interaction parameters and thus separation by spinodal decomposition or the effect of
positive heats of mixing [Bz(l - x)] for all systems. compositional fluctuations on possible precursory
The magnitude of the interaction parameter is actually reactions in these syst’ems.
slightly dependent upon temperature, but its sign It is evident’ that titanium-niobium has the largest
depends only upon the relative slopes of the phase two-phase region while titanium-molybdenum has t.he
boundaries of the two-phase region in each system. smallest. The beta phase is very stable beyond approx-
Although the regular solution approximation does imately 25 wt. % molybdenum. However. titanium-
not provide absolute values for the heats of mixing. vanadium and titanium-niobium alloys are unstable
it is sufficient,ly good for comparing different systems at even higher concentrations of solute. This is expec-
t.o determine the relative values and sign of the heats ted since titanium-molybdenum has the smallest
of mixing. heat of mixing. The results of the present thermo-
Kaufman and Bernstein have been able to predict dynamic calculations and experimental work on

the phase diagrams for a number of titanium alloys titanium-vanadium and titanium-niobium indicate
using this regular solution approximation. They have that probable phase diagrams, extended to tempera-
obtained values for the interaction parameter in the tures below those given in previously published
beta phase using instead the enthalpies of vaporiza- diagrams, or approximat,ely 600°C: should include
tion, the vol/g-at’om and the group number of the either a monotect,oid reaction (Fig. 1 la) or a metastable
two components. Values of this parameter and
calculated values of T,, the critical temperature given
by the expression ZRT, = B for the equiatomic
composition, are reproduced in Table 3. These values
TABLE 3. Calculated values of the interaction paramet,er (B)
and critical temperature (T,) (Reference 10)

Alloy system (cal,g:tom) (2,

Ti-Mo +1241 310

Ti-?ib +3125 781
Ti-V -+ 2659 665

of the interaction parameters are in qualitative agree-

ment with the present calculations with regard to
their sign and relative values.
It should be noted that these calculations find a
positive heat of mixing for titanium-molybdenum
alloys which is in disagreement with calculations by
Rostoker”*) and Rudmano2) who determined values
for the heat of mixing using a bond-energy concept.
/AG ”
HerG. a negative heat of mixing was found for titan-
ium-molybdenum while a positive value was obtained
for the titanium-niobium system. Further. X-ray
diffuse intensity measurements by Averbach and
Dupouy”s’ on titanium-molybdenum and by Rud-
mano2) on titanium-niobium allops showed ordering
and clustering tendencies. respectively.

Proposed phuse diayramv

The phase diagrams and precipitation behavior
of binary titanium alloys containing molybdenum,
niobium or vanadium can be rationalized in terms of T, V
the preceding calculation. A large, positive interaction PERCENT VANADIUM
parameter enlarges the two-phase (E + /?) regions. FIG. 11. Proposed phase diagram and free energy curves
The converse is true for negative values. The positive for the titanium-vanadium system illustrating increased
driving force for alpha formation in an equiatomic alloy
heats of mixing also suggest the possibility of phase undergoing phase separation.
586 ACTA METALLFRGICA, VOL. 18, 1970

miscibility gap within the (a f 8) phase region. occur. Precipitation is expect,ed to exhibit particu-
The titanium-niobium system, which has a higher larly slow kinetics because of the known very small
positive heat of mixing and closer-matching of diffusivity of niobium in this system.
atomic diameters than does titanium-vanadium, is Precipitation of alpha in titanium-50°b van-
more likely to undergo phase separation by spinodal adium can be rationalized in terms of compositiona.
decomposition. Ternary additions to titanium- fluctuations in this system however. In Fig. 11, where
niobium alloys which raise the critical temperature a phase diagram containing a monotectoid react’ion is
should encourage such coherent spinodal decom- assumed, the coherent, spinodal is depressed well
position in these alloys. within the (a + ,4) region because of strains arising
from the large atomic misfit. between tita,nium and
vanadium. Thus, the possibility of spinodal decom-
Phase separation and alpha precipitation
position can be eliminated at’ lower temperatures.
Experimental evidence for phase separation of t,he However, consideration of the free energy-composition
beta phase into solute-rich and solute-poor b.c.c. diagram at 5OO’C (Fig. llb) reveals that a composi-
phases was first reported in titanium-18 vanadium tional fluctuation in the 50°i vanadium alla)
and titanium-16 vanadium-2.5 aluminum by Troiano would appreciably increase the driving force for the
and Harmon.04) Splitting of beta phase reflections beta-to-alpha transformation and could induce pre-
was observed by X-ray diffraction when both alloys cipitation of the alpha phase. The different alpha
were aged at temperatures below 25O”C, which is precipitate morphology found at 350°C can be expec-
below the critical temperature of 850°C calculated by ted if it is below the monotectoid reaction tempera-
Krisement.@) Heats of evolution measurements in a ture. Oxygen contamination which can ext,end the
titanium-50 at. % vanadium alloy cooled from 950% range of alpha-phase stability is not a likely cause for
to 700°C did not exhibit any substantial effects the observed precipitation. The estimated distance of
attributable to phase separation. This was ascribed diffusion, given by the relatJionship 2 = 21/z, is
by Krisement to extremely slow precipitation not 3.2 x 10e3 cm after 30 days at 500°C using a value of
detectable with his microcalorimetric measurements. 1O-12cm2/sec for the diffusivity. Specimens for
He concluded also from measurements on several transmission electron microscopy were prepared by
titanium-vanadium alloys containing between 15 and electropolishing equal depths from both surfaces of an
19 at. % vanadium that all heat evolved during low aged sheet 0.38 mm in thickness. Further, aged
temperature aging could not be totally attributed to specimens exhibited no yield drop when deformed in
omega formation. Part of the heat evolved may tension.
therefore arise from phase separation and from alpha
precipitation. In agreement with the earlier work by
Omega phase formation and dissolution
Krisement, alpha precipitates were observed in aged
equiatomic titanium-vanadium alloys in the present It has been proposed(3J6) that the initial step in the
st,udy. formation of omega is phase separation in beta by
More recently, Blackburn and Williams(s) have spinodal decomposition. However, the observed
observed splitting of titanium-20 vanadium beta formation of omega in the titanium-molybdenum
reflections in electron diffraction. However, their system where an ordering tendency has been observed
results were not reproducible. Hickman finds no experimentally, argues against this suggestion invol-
X-ray evidence for phase separation in this system ving modulations in composition. It was also shown
and suggests that splitting of the beta reflections is in the previous section that the driving force for the
due to inhomogeneous precipitation occuring during beta-to-alpha transformation can be increased by a
very early stages. Nevertheless, a metsstable phase composition fluctuation in the beta phase of titanium-
diagram has been proposed(3) for titanium-vanadium vanadium alloys. In this way, phase separation in
which exhibits a metastable miscibility gap. They beta can act as a precursor to alpha precipitation.
further suggest that aging of titanium 20 wt.% Omega does not form in stable high-vanadium alloys.
molybdenum at low temperatures should result in The formation of omega, its composition, volume
phase separation. Contrary to this suggestion, a fraction dissolution and subsequent precipitation of
titanium-26 molybdenum alloy did not exhibit any the equilibrium alpha phase can be described using the
such tendency. Titanium-niobium alloys on the hypothetical free energy composition diagram given
other hand, should show phase separation. However, in Fig. 12. The relative positions of the free energy
this has not as yet been shown with certainty to curves for the beta. omega and alpha phases are
 KOCL ASD BREEDIS: PHASE TRASSFORJIXTIOSS IS TITASI~~M ALLOTS 587

There is a lack of agreement regarding the mode of

formation of equilibrium alpha from structures which
contain beta and omega phases. It is possible that
either alpha forms directly from omega or nuc1eates
in beta and subsequently grows to consume the omega
phase. Hickman(l5) suggests that equilibrium alpha
is formed by direct conversion of omega particles.
He reports that Rhodes (unpublished research)
has observed cube-shaped, alpha particles co-existing
with, and comparable in size to omega. part,icles.
Some particles also appear to have partially t,rans-
formed to alpha. From lattice parameter measurements
Hickman concludes that alpha does not form having
the expected equilibrium composition given b?
the phase diagram (approximatJely 1.5 at. % vana-
dium) but with a composition close to that of the
“pseudo-equilibrium” composition of omega with
beta, or approximately 13.6-14 at. “/‘, vanadium.
As aging proceeds. t,he alpha phase reject,s vanadium
and approaches the equilibrium composit,ion. Black-
burn et d.f31 on the ot,her hand. find that the rquilib-
rium alpha forms at the beta-omega interface in
titanium-vanadium alloys. For the cases of titanium-
niobium(41 and titanium-molybdenum~3~ alloys, the
---L---7
:/ : precipitation of the alpha phase seems to occur pro-
g=iJiz=
ferentially at, the grain boundaries b- a cellular or
f2 / *’ m_,_fi-. - - I/ heterogeneous nucleation process.
w
The observation in this investigation of alpha
/
precipitates between omega particles in alloys con-
taining beta and omega phases can be rat-ionalized as
follows. After long aging times at low temperatures,
the beta phase becomes saturated in vanadium as the
PERCENT VANADIUM large volume fraction of omega becomes richer in
FIG. 12. Hypothetical free energy curves (a), and (b) the
variation of solute content with disbance during the for-
titanium and finally precipitates the equilibrium
mation of alpha in an alloy containing beta and omega alpha-phase. It is seen from Fig. 12(b) that the vana-
phws.
dium conlposition of the beta matrix increases to
depicted for about 350°C. A het.a alloy having com- C,, as alpha forms. To maintain the vanadium con-
position C, partially transforms to omega of the same tent of the beta phase in equilibrium nit,b omega at
composition on quenching. During aging at 350°C. C,,, part of t’he otnega phase dissolves. This is shown
the omega precipitates become richer in titanium at in Fig. 12(b) where the solute distribution during
the expense of the beta phase until t’he omega and precipitation of the alpha phase is shown against
beta phases attain their equilibrium compositions, interparticle spacing. As the alpha precipitates grow,
CT, and C,,. In an alloy having cmposition C,, the additional omega particles dissolvr until finally the
omega can only form on aging. As seen from Fig. structure consists solely of alpha pax%icles in a, beta
12(a), the composition of omega in equilibrium with matrix.
beta is fixed at C, after aging irrespective of the
composition of the alloy studied. This agrees with the Stress-ivdu~ced vnarterisitc
experimental observations by Hickman(15) and Martensitic transformation occurs spontaneously
Bagariatskii and Sosova. (17) Further, application o at the M, temperature without an applied stress.
the lever rule to the tangent of the omega and betaf Since the ih!, temperature depends upon the net
curves shows that in agreement with Hickman, t*he driving force, it can be raised or lowered by the
. f
volume fraet.ion of omega. decreases with increasing application of an external stress.“*) In the former
alloy content. instance, the stress required for the initiation of the
588 ACTA METALLURGICA, VOL. 18, 1970

transformation increases with test temperatures above ation by spinodal decomposit,ion is possible in titan-
the M, temperature. The upper temperature limit ium-niobium alloys where the elastic st’rain arising
above which transformation cannot occur under from at,omic misfit is small.
stress is defined as the M, temperature. Kaufman’g) Compositional fluctuations in t,he beta phase of
has calculated T,, the temperature at which the beta titanium-vanadium alloys can lead to an increased
and alpha phases have equal free energy, for a number driving force for the formation of alpha.
of binary titanium systems. Assuming that T, and Omega formation is not associat’ed with the spon-
the M, .temperature follow the composition depend- taneous development of compositional modulations
ence of Al,, it is thermodynamically possible for in the beta phase.
stress-induced mart,ensite which is hexagonal close- The formation of alpha precipitates during aging
packed to form in titanium alloys containing between of a beta alloy containing omega can be rationalized
approximately 12-15 wt.% molybdenum or 15- in terms of its formation within the beta phase, per-
18 wt. % vanadium. While the iV, temperatures in haps at an interface. As the alpha precipitat,es gro\\..
these composition ranges lie well below room temper- omega dissolves to maintain the vanadium content
ature, as indicated bg Fig. 1, the M, lies around room of the beta phase.
temperature. Hexagonal close-packed, st,ress-induced martensite
Martensitic transformation during thinning can is thermodynamically possible. and has been found
complicate the identification of the stress-induced experimentally in molybdenum and vanadium con-
martensite structure. Metastable beta alloys having taining alloys.
lower solute contents are more prone to such trans-
ACKNOWLEDGMENT
formation. This complication was avoided in the
present study by select,ing a higher molybdenum- The authors wish to thank Professors Morris
containing alloy than studied previously,(3) and by Cohen and John Morral for valuable discussions.
minimizing the volume fraction of retained beta by a Miss Mirian Yoffa for experimental assistance, and to
large prior strain of approximately 15 per cent. acknowledge the support of this work by the Air
This investigation has shown that application of stress Force Materials Laboratory, Wright-Patterson Air
to metastable alloys induces transformation to Force Base, under contract Xo. AF 33(615)-3866.
martensite which is hexagonal close-packed in struc-
REFERENCES
ture. Mechanical t,wins do not occur readily in beta
1. J. M. SILCOCS. H. hf. DAVIES and H. K. HARDY, The Tech-
phase titanium-molybdenum alloys, even under anism of Phase Tnwsformations i)t Net&, p. 96. Institute
shock deformation conditions,(19) and nearly all of Metals (1956).
2. J. M. SILCOCK, Acta Met. 6, 481 (1958).
acicular structures ohserved can be ascribed to mar- 3. M. J. BLACKBURP; and J. C. WILLIAMS. Trans. metal1 SOC.
tensite. Finally, stress-induced martensite, hexagonal A.I.&f.E. 242. 2461 fl9681.
4. W. G. BRAMMER, JR’. and’ C. G. RHODES, Phil. Nag. 16,
close-packed in structure. was observed in both beta 447 (1967).
alloys. Earlier, only a b.c.c. or a t,etragonal structure 5. D. J. MAPKUTH, H. R. OGDEX and R. I. JAFFEE. Thtb
Effects of Alloyinfi Elements m Titanium, DMIC R&ort
was reported. This latter structure must’ result from 136, Batt,elle Memohal Instltutc (1960-61).
transformation during thinning. . 6. G. S. UPADHYAYA and A. D. MVQOILLAX. Trans. mstnll.
Sot. A.Z.Jf.E. 224, 1290 (196”).
7. L. D. JAFFEE, J. Ndals 8. 861 ( 1956).
SUMMARY
8. 0. KRISEDCENT, Z. Ndallk. 52, 695 (i!Wl).
Thermodynamic considerations. employing t,he 9. L. KAUF~IAS, Acta Net. 7, 575 (1959).
10. L. KAUF,VAX and H. BEKNSTEIS. Ww Srir~~rc, Technolopy
regular solution a~~~)rosimatio~l. suggest a tendenc)
for phase separation of the beta phase in binar)
titanium alloys containing niobium, vanadium and 19. 1’. S. RWXXAS. Artn Xllct. 12, 13SI (1964).
molybdenum in this ortlcr. C’ompletc~ phase diagrams 13. I). L. ~VERBACH, .I. 11. I)I~Po~-Y. .4c/rr Alrt. 9, 555 (1961).
14. A. R. THOIANO and EL L. HAIWOK. .IH.. Y’rn~. dr,t.
for the two former systems sl~ould colltnin either a Ser. Xetalu 53, 43 ( 1961).
monotectoid reaction or a metastablc spinodal. 15. 13. S. HICKJIAN. ./. Inst. .Il(,tds 96, 330 ( IMH).
16. I. .I. BAGAIIIATSKII, Q. .I. SWOYA. 1’. V. TAGI~NOVA,
depending upon the magnitude of the heat of mixing. Socir/ Php. I)okl. 3, 1041 (1958).
On the basis of experimental observations. it, is 17. I. A. BAGARL~TSKII and (:. I. Sosova. I’kysicn ?)fet&
J~e/azlo~r. 13, 92 (196”).
proposed that a complctc~ phase diagram for the 18. L. KACFMAS and 11. (‘omz~. I’roq. X&Z Phyn. 7, 165
titanium-vanadium aud titanium-niobilIln system (1958).
19. M. Ii. Ko~.L and .J. F. BREEDIS. The S’rie~c~. Trchnolo!pl
should include a monotectoid reaction. Phase separ- ant2 Application of Titonirrur, p. iO.5. l’wgamon (1969).