Conversion of titanium hydride to titanium nitride during mechanical milling

Article abstract of DOI:10.1103/PhysRevB.55.14810

The response of a stable titanium hydride to severe mechanical milling treatments at room temperature was studied in different gaseous environments using different milling media. Structural details were monitored by Reitveld refinements using x-ray-diffraction data and hydrogen compositions were measured by decomposition at 850 °C in a constant volume system. Milling in an argon atmosphere produces only nanocrystalline TiH_x with a reduction in hydrogen content proportional to the milling time. Milling in steel and Co bonded WC vials in air gave a more rapid loss of H and simultaneous formation of TiN. The lattice parameters of both the TiH_x and the Ti formed by decomposition support a mechanism based on solid solution of N in the TiH_x followed by partition into two phases. This resembles a chemically enhanced version of the solid-solution pumping mechanism previously described for the nitrogenation of pure Ti milled in air. Experiments with different milling media and minor additions of Fe and or Co suggest that 3d transition metal is necessary for the nitrogenation to occur, most likely by promoting dissociative chemisorption of N₂.

Full text of DOI:10.1103/PhysRevB.55.14810

PHYSICAL REVIEW B
VOLUME 55, NUMBER 22
1 JUNE 1997-II
Conversion of titanium hydride to titanium nitride during mechanical milling
Heng Zhang and Erich H. Kisi
Department of Mechanical Engineering, The University of Newcastle, Callaghan, New South Wales 2308, Australia
͑Received 2 December 1996͒
The response of a stable titanium hydride to severe mechanical milling treatments at room temperature was
studied in different gaseous environments using different milling media. Structural details were monitored by
Reitveld refinements using x-ray-diffraction data and hydrogen compositions were measured by decomposition
at 850 °C in a constant volume system. Milling in an argon atmosphere produces only nanocrystalline
TiHx with a reduction in hydrogen content proportional to the milling time. Milling in steel and Co bonded WC
vials in air gave a more rapid loss of H and simultaneous formation of TiN. The lattice parameters of both the
TiHx and the Ti formed by decomposition support a mechanism based on solid solution of N in the TiHx
followed by partition into two phases. This resembles a chemically enhanced version of the ‘‘solid-solution
pumping’’ mechanism previously described for the nitrogenation of pure Ti milled in air. Experiments with
different milling media and minor additions of Fe and or Co suggest that 3d transition metal is necessary for
the nitrogenation to occur, most likely by promoting dissociative chemisorption of N2.
͓S0163-1829͑97͒01121-1͔
I. INTRODUCTION
pects the hydride to be the least stable. During conventional
hydriding, TiHx can tolerate a wide band of hydrogen con-
centrations without undergoing structural phase changes
from the body-centered tetragonal TiH2 structure. There is
however a temperature-induced transition to the cubic CaF2
structure at 310 K.30 All these features lead us to expect a
high likelihood that TiH2 would undergo interesting changes
during milling.
Solid-state phase transitions in intermetallic compounds
during high-energy ball milling have generated considerable
interest.1,2 An initially homogeneous intermetallic compound
can be transformed to an amorphous state ͓YCo,3 NiZr,4
NiTi,5 ZrAl,6 CoZr,7 NiSn,8 CoTi,9 NbSn,10 CoGe, and GeAl
͑Ref. 11͔͒; a solid solution ͓V3Ga and Nb3Au ͑Refs. 12 and
13͔͒; or undergo a crystal structure transformation ͓e.g.,
Co3Sn2 and Ni3Sn2 ͑Refs. 14 and 15͔͒. In addition, atomic
disorder during the early stages of milling has been observed
in many ordered compounds with L12-type structure
II. EXPERIMENTAL
The starting material for these experiments was TiH1.73
powder ͑Research Organic/Inorganic Chemical Corporation,
Sun Valley, California 91352͒ with particle size about 5–8
␮m (Ϫ325 mesh). The TiH1.73 powders were milled in a
Shaker mixer/mill ͑Spex 8000͒ with an operating cycle of 45
min on and 15 min off. The majority of the ball milling was
performed in a hardened steel vial and balls under an air or
argon atmosphere. The argon and air atmospheres were ob-
tained by charging the powders into a vial in an argon-filled
glove box or just in air, respectively. Approximately 5 g of
powder was sealed by an O ring in a vial and the initial
weight ratio of balls to powders was 3:1. After different mill-
ing times, the vial was opened in a glove box under argon
͑for an argon atmosphere͒ or directly in air ͑for an air atmo-
sphere͒ and a small amount of powder was taken out for
x-ray diffraction ͑XRD͒. A similar milling procedure was
also carried out in zirconium dioxide and tungsten carbide
vial and balls operated in air, in order to identify the effect of
the milling media.
Ni Al,16 Ni3Ge,6 Ni3Si ͑Ref. 17͔͒; B2-type structure
͓
3
͓AlRu,18 CoAl,19 CoGa,21 and CoZr ͑Ref. 21͔͒; B8-type
structure ͓Mn3Sn2, Fe3Sn2,22,23 Co3Sn2, Ni3Sn2,14,15
Co2Ge,24 and Co2Si ͑Ref. 25͔͒; and the A15 structure
Nb Au ͑Ref. 26͒ and Nb Sn ͑Ref. 27͔͒. Investigation of the
͓
3
3
milling of intermetallic compounds has lead to the synthesis
of metastable phases or states with special characteristics un-
able to be obtained by other methods.
In addition to the above metal-metal or metal-metalloid
systems, there is another group of compounds, the metal-gas
systems such as oxides, nitrides, and hydrides. While the
nitriding of milled pure metals or alloys has been studied,
there are no reports of the effects of milling on stable nitrides
or hydrides. When compared to intermetallic compounds,
these metal-gas compounds have an additional degree of
freedom. That is, gas atoms can be desorbed or absorbed
during the milling process thereby causing the system to
have a variable composition and mass.
In considering oxides, nitrides and hydrides, it should be
remembered that often the hydride phase will have the low-
est enthalpy of formation. A perfect illustration is the system
studied here. Stoichiometric TiH2 has a formation enthalpy
of about Ϫ14.96 kJ/mol,28 whereas TiN and TiO have
Ϫ336.6 and Ϫ542.9 kJ/mol, respectively.29
A Sieverts-type gas-solid reaction apparatus was used to
measure the amount of dehydrogenation induced by milling,
and the stability of the milled products. The ultimate vacuum
in this system is about 4ϫ10Ϫ4 Pa and the sensitivity to gas
pressure change is 700 Pa. The hydrogen content of the pow-
ders was calculated from the pressure rise during heating to
850 °C in an initially evacuated chamber of known volume.
The structure information was monitored by using a Phil-
If we take the size of the enthalpy of formation as a first-
order indication of the stability of the compound, one ex-
0163-1829/97/55͑22͒/14810͑8͒/$10.00
55
14 810
© 1997 The American Physical Society

55
CONVERSION OF TITANIUM HYDRIDE TO TITANIUM . . .
14 811
ips PW1700 x-ray diffractometer ͑XRD͒ with Cu K␣ radia-
tion. Rietveld refinement analysis based on the x-ray-
diffraction data was used to assist with phase identification
and the refinement of lattice parameters. The microstructure
of the milled powders was observed by scanning electron
microscopy ͑SEM͒ ͑JOEL JSM-840͒ with 15-kV accelerat-
ing voltage. Powders for SEM examination were mounted in
epoxy resin and polished with 1 ␮m diamond to give a cross-
sectional view of agglomerates.
III. RESULTS
The x-ray diffraction patterns at different stages of milling
are shown in Fig. 1. Figure 1͑a͒ is for powders milled in
argon and Fig. 1͑b͒ for milling in air. Figure 1͑a͒ shows
a pattern typical of the original body-centered-tetragonal
structure ͑space group I4/mmm͒ of titanium hydride.
The c/a ratio is about 1.3925 ͓a0ϭ3.1596(6) and c0
ϭ4.3996(9) Å͔ based on Rietveld refinement. The lattice
parameters for the initial TiH1.73 are less than those of the
stoichiometric titanium dihydride, a0ϭ3.202 and c0
ϭ4.279 Å ͑Ref. 30͒ because of the lower hydrogen content.
After milling for 5 h, all the diffraction peaks were broad-
ened, and the overlapping of peaks corresponding to ͑222͒,
͑311͒, ͑113͒ planes can be observed. It is no longer immedi-
ately clear if the structure is still body-centered-tetragonal, or
if it has converted to the high-temperature cubic form ͑see
below and Sec. IV A͒. Further development of the broaden-
ing of all the diffraction peaks can be seen after milling for
15 and 30 h. For the sample milled for 30 h, a set of weak
peaks corresponding to a new phase can be distinguished in
addition to the peaks corresponding to the titanium hydride.
In the light of the results below for milling in air, we believe
these peaks resulted from light leakage of air into the vial.
The considerable broadening of all the diffraction peaks is
primarily due to the refinement of crystallite size with a
small contribution from internal stress induced by severe me-
chanical deformation.
The structural evolution of the titanium hydride milled in
air is quite different ͓Fig. 1͑b͔͒. Again, broadening of all the
Bragg diffraction peaks appears after just 5 h milling. The
difference occurs when a new set of diffraction peaks at ap-
proximately dϭ2.12, 1.50, 1.29, and 1.22 Å become promi-
nent after milling for 15 h. After milling for 30 h, all the
peaks from titanium hydride have disappeared, and were re-
placed by a new phase with the NaCl structure; either tita-
nium nitride or titanium oxide. The new phase keeps the
same structure after a further 20 h milling ͑except for some
iron impurity from the milling medium͒ showing that it pos-
sesses good stability ͓Fig. 1͑b͔͒.
Figure 2 shows the crystallite size as a function of milling
time. It was estimated using the Scherrer equation and Ri-
etveld refinement31 results after correction for the instrument
profile. It should be noted that when measured in this way,
‘‘crystallite size’’ refers to the mean size of coherently dif-
fracting crystal domains and does not indicate how the crys-
tallites are assembled into particles nor how the particles are
assembled into agglomerates. The results indicate that the
crystallite size for the titanium hydride decreases very
quickly during the early stages of milling. After milling for
30 h in argon the crystallite size was reduced to only 45 Å.
FIG. 1. XRD patterns for TiH1.73 milled under argon ͑a͒ and air
atmosphere ͑b͒ vs milling time.
Milling in air has a very similar effect on the particle size of
the titanium hydride ͑Fig. 2͒ except that the formation of the
new phase makes the determination unreliable beyond 15 h.
The crystallite size of the new phase increases with further
milling even when it comprises 100% of the powder. Be-

14 812
HENG ZHANG AND ERICH H. KISI
55
FIG. 2. Crystallite size of titanium hydride milled under argon
᭺ and air ϩ atmosphere, and the new phase ᮀ vs milling time.
FIG. 4. The atomic ratio of hydrogen to titanium H/Ti as a
function of milling time for milling in argon ᮀ and air ࡗ measured
by decomposition at 850 °C.
cause our crystallite size measurements refer to coherently
scattering ͑near perfect͒ domains, we interpret the size in-
crease as indicating recrystallisation of adjoining domains to
form larger crystallites of the new phase.
͑measured by decomposition at 850 °C͒ are presented in Fig.
4. The structure was tested by XRD after dehydrogenation
and the results are shown in Fig. 5. For the case of milling in
argon, the residual hydrogen composition TiHx decreases
The microstructures generated during milling in the steel
vial in air can be seen from the scanning electron microscope
͑SEM͒ images presented in Fig. 3. The complicated and
heavy deformation during milling causes the hydride powder
to be crushed into fine particles. Simultaneously, the fine
particles are agglomerated into aggregates of quite large size.
These two processes lead to a quite inhomogenous particle
size distribution. Fine particles of less than 0.5 ␮m coexist
with agglomerates up to 500 ␮m. As milling continues, the
fraction of very fine particles increases, however, the size of
the larger agglomerates also increases. We believe that the
milling is quite dynamic with agglomerates fragmenting and
reforming continually.
Results for the hydrogen content of the milled powders
FIG. 3. Microstructure of titanium hydride and the new phase
during the milling process ͑a͒, ͑b͒ 15 h, ͑c͒ 30 h, and ͑d͒ 50 h.
FIG. 5. XRD patterns for the samples dehydrogened at 850 °C
after different milling times in air.

55
CONVERSION OF TITANIUM HYDRIDE TO TITANIUM . . .
14 813
FIG. 7. XRD patterns for the mixture of (TiH1.73
)
98M2 ͑wt %͒
MϭFe for ͑a͒, and Co for ͑b͔͒ milled in a ZrO vial under an air
͓
2
atmosphere.
results for milling in the WC vial are very similar to those
for the steel vial; all the hydride transformed to a new phase
with the NaCl structure after 30 h milling. However, the
situation using the ZrO2 vial is very different, with only
broadening of the hydride diffraction peaks observed after 30
h milling. The difference of the structural evolution using the
different vials ͑Fe, WC, and ZrO2͒ indicates that the milling
medium is a very important factor during the milling of tita-
nium hydride.
It was noted that the WC vial material is bonded by co-
balt. The presence of a transition metal element ͑Fe, Co͒ may
play a special role in the transition. Two mixtures of titanium
hydride with 2% ͑wt͒ iron, or cobalt powder were milled
with the ZrO2 vial under identical experimental conditions as
the pure hydride. The results are shown in Fig. 7 and give
strong support for the above hypothesis. In both cases most
of the hydride transformed to a new phase with the NaCl
structure.
FIG. 6. XRD patterns for TiH1.73 milled in a WC vial for ͑a͒ 5 h,
͑b͒ 15 h, ͑c͒ 30 h, and ͑d͒ in a ZrO2 vial for 30 h in air.
from xϭ1.73 initially to xϭ1.480, 1.162, and 0.893 after
milling for 5, 15, and 30 h milling, respectively. An interest-
ing phenomenon is that the hydride still appears to display a
bct structure although the ratio of H/Ti is only 0.893 ͑see
discussion͒. For milling in air, the dehydrogenation rate is
much faster than that under an argon atmosphere. The ratio
of H/Ti decreases to 0.316 and 0.105 after milling 30 and 50
h, respectively. For the latter two samples, there was still
residual hydrogen although the sample had transformed to a
new phase ͓Fig. 1͑b͔͒.
Examination of the XRD patterns shows that dehydroge-
nation of the starting material TiH1.73 at 850 °C results in
pure Ti ͑hexagonal, P63 /mmc͒ as expected.32 Samples
milled for 5 to 30 h under an argon atmosphere also show
only the titanium XRD pattern after dehydrogenation.
However, the results are quite different for the samples
milled in air ͑see Fig. 5͒. The sample milled for 5 h shows
the titanium structure. After 15 h milling, in addition to the
Ti pattern, two weak diffraction peaks are seen at dϭ2.12
and 2.42 Å. These correspond to a new phase with the NaCl
structure. For the sample milled for 30 h, only sharp lines
corresponding to the new phase with the NaCl crystal struc-
ture are visible. The lines have sharpened because of the
coarsening of the crystallite size and the elimination of de-
fects. The ability to survive heating to 850 °C in high
vacuum provides extra evidence for the stability of the new
phase.
IV. DISCUSSION
A. Phase analysis for the milled products
It was apparent in Sec. III that milling under argon causes
TiHx to partially decompose. In addition, the crystallite size
is rendered extremely fine. Visual inspection of the diffrac-
tion patterns ͓Fig. 1͑a͔͒ is not sufficient to ascertain whether
the milled powders have returned to the bct low-temperature
form or remained in the high-temperature fcc phase that ex-
ists at the average milling temperature (ϳ323 K). Rietveld
analyses based on the fcc model return slightly worse R val-
ues than for the bct model. As an example, RBϭ7.85% for
the fcc model and RBϭ6.33% for the bct model applied to
the XRD data for the sample milled for 30 h in argon. With
Figure 6 presents the XRD patterns for the titanium hy-
dride milled in air with WC and ZrO2 vials, respectively. The

14 814
HENG ZHANG AND ERICH H. KISI
55
TABLE I. Results of Rietveld refinements for TiH1.73 at milled
and annealed states. A: TiH1.73 milled 30 h; B: TiH1.73 an-
nealed at 850 °C after milled 30 h.
Sample
A
B
Model
TiN
Parameters
a ͑Å͒
BTi
4.230͑3͒
1.0͑1͒
4.2409͑3͒
0.4͑1͒
BN
3.0͑3͒
0.2͑2͒
r ͑Å͒
(RB %)
GOF
70.0
4.33
1.37
761
3.28
1.59
TiO
Bi fixed
Parameters
a ͑Å͒
BTi
4.232͑3͒
1.0͑1͒
3.0͑3͒
9.23
4.2408͑4͒
0.4͑2͒
0.2͑2͒
9.15
FIG. 8. Rietveld refinement results for samples milled for 30 h
BN
(RB %)
GOF
in air ͑a͒ and subsequently annealed at 850 °C ͑b͒. The XRD data
are given as ϩ signs and the difference plot and calculated intensi-
ties are shown by solid lines.
1.56
1.76
TiO
Bi variable
Parameters
a ͑Å͒
BTi
so few reflections and such broadened peaks, we must regard
this improvement as supporting the bct model, but not as
conclusive evidence. For example, the Hamilton test39 ap-
plied to these RB values indicates that we may have about
80% confidence in rejecting the fcc model.
4.228͑3͒
1.22͑2͒
6.7͑4͒
6.09
4.2408͑3͒
0.6͑1͒
2.92͑3͒
3.29
BN
(RB %)
GOF
1.43
1.59
The powders milled in air present a more complex situa-
tion. Milling results in decomposition of the hydride and
eventual reaction of the milled product with air to form ei-
ther TiN or TiO. It is not immediately apparent which is
formed since their XRD patterns and lattice parameters are
similar. Both can be nonstoichiometric with a corresponding
range of lattice parameters and the Ti L lines overlap the
nitrogen and oxygen K lines in energy-dispersive spectros-
copy spectra.
that TiN forms in preference to TiO. The increase in lattice
parameter after vacuum annealing at 850 °C to 4.2409͑3͒ Å
is presumably due to the annihilation of lattice defects such
as Ti vacancies. The lattice parameter of TiNx is known to
decrease as x is varied from 1.39,40
B. Milling under argon
Previous work on pure Ti milled in air38 provided consid-
erable evidence that the compound formed was TiN, as sum-
marized below:
The results obtained from milling TiH1.73 under argon
provide a reference which helps us understand the more
complex behavior during milling in air. There are two pri-
mary effects here; rapid crystallite size reduction ͑Fig. 2͒ and
partial dehydriding ͑Fig. 4͒. In addition, there are more
subtle effects such as a reduction in the unit-cell volume and
the c/a ratio. This is summarized in Table II͑a͒. Both are
believed to result from loss of hydrogen during the milling
process. Titanium hydride is very stable since, under normal
circumstances, temperatures exceeding 800 °C are required
to cause total decomposition. We take this to indicate either
that the milling process is so severe, it causes very localized
temperature excursions approaching 800 °C or that the com-
bined mechanical and thermal excitation locally supplies an
equivalent amount of energy to the hydride.
͑i͒ The lattice parameter of the milled product ͓4.230͑3͒
Å͔ is far closer to that of TiN ͑4.2400 Å͒,33 than that of
stoichiometric TiO ͑4.174 Å͒.34
͑ii͒ The effect of nonstoichiometry is to reduce the lattice
parameter due to the formation of vacant Ti sites. The lattice
parameter of TiOx may be described as35
aϭ4.23Ϫ0.05167x,
which can clearly not attain a value of 4.230 for nonzero
values of x since the minimum oxygen composition for
TiOx can be about 0.536.36
͑iii͒ Rietveld analyses based on the XRD data show that
͑a͒ the agreement index RB is poor and,
͑b͒ unrealistic thermal parameters result when a TiO
model is used in the analysis. Further explanation is
given in Ref. 37.
C. Hydrogen desorption and nitrogen absorption
during milling in air
Before discussing the results obtained by milling TiH1.73
in air, we pause to comment briefly on the milling environ-
ment. The powder is subjected to repeated ball/vial and ball/
ball collisions. As shown by the SEM micrographs, there is
coalescence of finer particles into large agglomerates. Simul-
taneously, the size of the ‘‘perfect’’ crystallites is progres-
sively reduced and parts of the material are reacting with
nitrogen from the air. It is difficult to form a general model
͑iv͒ Heating under high vacuum leads to a lattice param-
eter very close to that of stoichiometric TiN.
͑v͒ Laser Raman spectroscopy and x-ray photoemission
spectroscopy also supported these conclusions.
Similar tests ͑i–iv͒ were applied to the milled product in
these experiments ͑Fig. 8 and Table I͒ leading us to conclude

55
CONVERSION OF TITANIUM HYDRIDE TO TITANIUM . . .
14 815
TABLE II. Unit-cell parameters from the Rietveld Refinement for TiH1.73 milled under argon, air atmo-
sphere, and annealed at 850 °C after milling. TiH1A , TiH2A , and TiH3A are TiH1.73 milled 5, 15, and 30 h
under an argon atmosphere. TiH0, TiH1, and TiH2 are the TiH1.73 milled 0, 5, and 15 h in air. TiH0.8
,
TiH18, and TiH28 are samples annealed at 850 °C after milling for 0, 5, and 15 h in air. The parameters are
based on the bct TiH2 structure for the milled series and hcp Ti for the annealed series.
Parameter
a ͑Å͒
c ͑Å͒
c/a
Vunit (Å3)
Sample
͑a͒
Milled
in Ar
TiH1A
TiH2A
TiH3A
3.714͑3͒
3.144͑6͒
3.200͑4͒
4.382͑1͒
4.324͑2͒
4.198͑7͒
1.3806
1.3535
1.3119
44.15
44.13
43.00
͑b͒
Milled
in air
TiH0
TiH1
TiH2
3.1596͑6͒
3.175͑3͒
3.191͑3͒
4.3996͑9͒
4.366͑4͒
4.356͑6͒
1.3925
1.3751
1.3651
43.92
44.01
44.35
͑c͒
Vacuum
annealed
TiH08
TiH18
TiH28
2.938͑3͒
2.9660͑4͒
2.9738͑6͒
4.707͑3͒
4.7743͑4͒
4.7937͑6͒
1.6021
1.6099
1.6121
35.19
36.38
36.71
for such a chaotic nonequilibrium environment. However,
there is a remarkable degree of homogeneity in the final
milled powders which can be revealed by vacuum annealing
the powders as outlined below.
treatment Ti increases with milling time. After 15 h of mill-
ing, there is also a small amount of TiN precipitated ͑Fig. 5͒.
The precipitation and growth of TiN is the final stage of
the process. After an apparent incubation time of nearly 15 h,
it proceeds to completion within a further 15 h ͑30 h total
milling time͒ and is insensitive to further milling ͓Fig. 1͑b͔͒.
It appears that the nitrogenation occurs by three overlap-
ping processes; H desorption, N absorption, and TiN nucle-
ation and growth. Since the milling vial and its contents
warms up to approximately 323 K during use, we expect the
hydride to quickly transform to the cubic high-temperature
form. All subsequent events take place in this form.
It is quite clear from Fig. 4 that there is enhanced desorp-
tion of hydrogen when the hydride is milled in the presence
of air. We believe this is due to the absorption of N in the
early stages of milling. Evidence for nitrogen absorption
throughout the milling process comes firstly from an increase
in the unit-cell volume, despite the loss of some H. Table
II͑b͒ shows this effect and it may be contrasted with the
reduction in cell volume that occurs during milling in argon
͓Table II͑a͔͒. The unit cell of TiN is smaller than that of
TiH1.73 ͑in the pseudo-cubic face-centered setting͒, so only
by combined occupancy by H and N may we understand a
cell expansion.
During the transition TiHx to TiN, the tetrahedrally coor-
dinated H is replaced by the octahedrally coordinated N at-
oms. In this sense, it might be expected that the transition
occurs in a smooth fashion across the composition range
TiHxNy ͑x from 1.7→0, y from 0→1͒. The XRD traces tend
to contradict this expectation. In Fig. 1͑b͒, there are defi-
nitely two sets of diffraction peaks indicating that there are
two populations of crystallites with different average struc-
tures, i.e., the solubility limit of N in TiHx is fairly low
before the hydride becomes unstable to TiN.
D. The mechanism of dehydrogenation and nitrogenation
during milling
Standard reaction mechanisms of anion diffusion through
a thick film surface layer are not appropriate to either the low
average temperature of the milling environment or the rapid
and repeated mechanical deformation of the powders in this
process.
The previous section presents evidence for the formation
of a solid solution of N in TiHx as an intermediate step to the
nucleation of TiN. This is remarkably similar to observations
of the nitrogenation of pure Ti by milling in air.38 In that
case, it was proposed that a ‘‘solid-solution pumping’’
mechanism operates. The mechanism may be summarized as
proceeding in four steps.
͑i͒ A thermal spike occurs very locally in the region of a
collision event. During the temperature spike, N is absorbed
into ␣-Ti ͑in that case͒ to form a solid solution, which is
retained by the rapid decay of the thermal spike, trapping N
in the Ti.
͑ii͒ Successive collisions raise the N content to a level
close to the solubility limit.
͑iii͒ TiN nucleates homogenously within the powder par-
ticles as the solubility limit is exceeded.
͑iv͒ Because the solubility of N falls as the temperature
falls, the solid solution becomes supersaturated each time it
‘‘cools’’ after a collision. Excess N diffuses from the solid
solution causing growth of the TiN precipitates. In a sense,
the solid solution ‘‘pumps’’ N into the precipitates.
It would appear that a similar mechanism operates here,
with the exception that the starting material is TiH1.73 instead
of Ti, and the desorption of H has to be incorporated into the
model.
Further evidence comes from samples milled for 5 and 15
h followed by vacuum annealing at 850 °C ͓Table II͑c͔͒. We
know that such treatment causes removal of all of the H and
milling induced effects. Any residual effect on the unit-cell
parameters of the resulting Ti is due to absorbed N. From
Table II͑c͒ it can be seen that the cell volume of the post heat

14 816
HENG ZHANG AND ERICH H. KISI
55
The Ti—N bond is largely covalent so firstly we expect a
large local perturbation to the charge distribution of the Ti
atoms in the vicinity of a N atom. Secondly, the H atoms
now have their third NN and fifth NN H-H interactions lo-
cally screened by the N atom. The third effect of a N atom in
TiH2 is the direct N-H interaction since they are now in close
NNN positions. The N assisted dehydrogenation is likely to
have been assisted by ͑i͒ the starting material being nonsto-
ichiometric and ͑ii͒ initial H loss during the very early stages
of milling.
The net effect is to promote more rapid decomposition of
the hydride. It should be noted that the rate of TiN formation
is very similar to that for pure Ti milled in air, and is prob-
ably limited more by the supply of N2 reaction gas than
internal gas-metal reaction processes.
In previous work on the milling of Ti,38 it was noted that
the presence of Fe in the milling vial and balls was necessary
to the formation of TiN. No TiN was observed in powders
milled in a ZrO2 vial until a small amount of Fe powder was
introduced into the milling charge.
These observations have been reinforced here. TiH1.73
milled in air yielded only nanocrystalline TiHx ͑Fig. 6͒ when
a ZrO2 vial was used. However in a Co bonded WC vial,
again the product was TiN.
However, TiH1.73 mixed with 2 wt. % of Fe or Co and
milled in the ZrO2 vial yields TiN ͑Fig. 7͒. We conclude that
the presence of minor quantities of a group VIII transition
metal is necessary for the nitrogenation to proceed. In previ-
ous work38 we speculated that this may be due either to
catalysis of dissociative chemisorption of N2 on the Fe or Co
during collisions. An alternative suggestion is that the tran-
sition metals serve as O sinks which reduce the oxygen par-
tial pressure, making nitrogenation the favored reaction.
FIG. 9. A ͓001͔ projection of the Ti atoms in the TiH2 structure
͑with CaF2 structure͒ at zϭ1/2: ᭹, and the coordinating H atoms at
zϭ1/4 and zϭ3/4: •. The octahedral sites where interstitial N is
absorbed are also shown: ᭺.
It is clear that milling alone is enough to cause consider-
able H desorption ͑Fig. 4͒, most likely through large, very
localized temperature excursions ͑the hydride is normally
relatively stable to Ͼ700 °C͒. It is also clear that the absorp-
tion of N into TiHx leads to an enhancement of the H de-
sorption rate. This is quite likely due to a coupling of the
thermo/mechanical effects above with chemical and struc-
tural effects.
The nitride is for more chemically stable so we expect a
large thermodynamic driving force favoring TiN formation.
Some insight may be gained from a comparison of the bond
strengths, 557.4 kJ/mol ͑Ref. 42͒ for Ti—N and 159 kJ/mol
͑Ref. 41͒ for TivH bonds, respectively.
V. CONCLUSIONS
In addition we must consider the local environment of a N
atom absorbed into titanium hydride. Referring to Fig. 9, the
Ti atoms form an fcc framework within which the H atoms
occupy all of the tetrahedral holes in stoichiometric
TiH2 . TiN has the same Ti arrangement, but N occupies
the octahedral holes. In the hydride (TiH2) the octahedral
void has a radius ͑defined by near-neighbor Ti atoms͒ of
approximately 0.76 Å compared with a nominal radius of
0.75 Å for N.
Ball milling of titanium hydride under an argon atmo-
sphere induces progressive dehydrogenation. Nanocrystalline
TiH0.89 was obtained after 30 h milling, while milling in air
causes simultaneous dehydrogenation and nitrogenation of
the titanium hydride. After 30 h of milling, nearly stoichio-
metric TiN with the NaCl crystal structure and good stability
was produced.
The nitrogenation process in titanium hydride is quite
similar to that for pure Ti milled in air, but completely dif-
ferent from traditional gas-solid reaction processes. A
mechanical-chemical nitrogenation mechanism was pro-
posed to explain the unusual dehydrogenation and nitrogena-
tion induced by ball milling. Formation of a saturated-solid
solution of nitrogen by ‘‘pumping’’ plays a critical part in
nucleation and growth of the nitride during the nitrogenation
process. This work further confirms that 3d transition metals,
Fe and Co from the milling medium appear to play an im-
portant role in the nitrogenation process.
The H atoms in TiH2 each have four nearest-neighbor
͑NN͒ Ti atoms along 1/4͗111͘, six next-nearest-neighbor
͑NNN͒ H atoms along 1/2͗100͘, 12 third-NN H atoms along
1/2͗110͘, and eight fifth-NN H atoms along 1/2͗111͘ as
largely unobstructed neighbors ͑the fourth-NN are partially
screened Ti atoms along 1/4͗311͒͘. The solution of a N atom
into TiH2 therefore requires N to occupy a site with six NNN
Ti atoms in an octahedral coordination shell and eight NN H
atoms along 1/4͗111͘ ͑between atom centers although the Ti
atom is much larger and will exert the greater influence͒.

55
CONVERSION OF TITANIUM HYDRIDE TO TITANIUM . . .
14 817
1 C. C. Koch, Mater. Trans. JIM. 36, 85 ͑1995͒.
20 L. M. Di and H. Bakker, and T. Tamminga, Phys. Rev. B 44,
2444 ͑1991͒.
2 A. W. Weeber and H. Bakker, Physica B 153, 93 ͑1988͒.
3 A. Y. Yemakov, Y. Y. Yurchikov, and V. A. Barinov, Fiz. Met.
Metallovd. 52, 1184 ͑1981͒.
21 G. F. Zhou and H. Bakker, Physica B 211, 134 ͑1995͒.
22 G. F. Zhou and H. Bakker, Phys. Rev. B 48, 7672 ͑1993͒.
23 G. F. Zhou and H. Bakker, Phys. Rev. B 49, 12 507 ͑1994͒.
24 G. F. Zhou and H. Bakker, Phys. Rev. B 48, 13 383 ͑1993͒.
25 G. F. Zhou and H. Bakker, J. Phys. Condens. Matter 6, 4043
͑1994͒.
4 R. B. Schwarz and C. C. Koch, Appl. Phys. Lett. 49, 146 ͑1986͒.
5 D. L. Beke, H. Bakker, and P. I. Loeff, J. Phys. 14, c4-63 ͑1990͒.
6 T. Benameur and A. R. Yavarari, J. Mater. Res. 7, 2971 ͑1992͒.
7 A. Mehrtens, G. Von Minnigerode, and K. Samwer, Z. Phys. B
83, 55 ͑1991͒.
26 L. M. Di and H. Bakker, J. Appl. Phys. 71, 5650 ͑1992͒.
27 L. M. Di, P. I. Loeff, and H. Bakker, J. Less-Common Met. 168,
183 ͑1991͒.
8 T. T. Tiainen and R. B. Schwarz, J. Less-Common Met. 140, 99
͑1988͒.
28 T. R. P. Gibb, Jr J. J. Mscharry, and R. W. Bragdon, J. Amer.
Chem. Soc. 73, 17051 ͑1951͒.
9 Y. Seki and W. L. Johnson, in Solid State Powder Processing,
edited by A. H. Clayer and J. J. de Barbasillo ͑TMS, Warren-
dale, PA, 1990͒, p. 287.
29 Eric A. Brandes, Smithells Metals Reference Book, 6th ed. ͑But-
terworths, London, 1986͒, pp. 8-24, 8-27.
10 Y. S. Cho and C. C. Koch, Mater. Sci. Eng. A 141, 139 ͑1991͒.
11 G. F. Zhou and H. Bakker, Phys. Rev. Lett. 72, 2290; 73, 344
͑1994͒.
30 H. L. Yakel, Acta Crystallogr. 11, 46 ͑1958͒.
31 R. J. Hill and C. J. Howard ͑unpublished͒.
32 A. D. McQuilan, Proc. R. Soc. London 204, 309 ͑1951͒.
12 L. M. Di and H. Bakker, J. Phys. Condens. Matter 3, 3427 ͑1991͒.
13 H. Bakker and G. F. Zhou, Mater. Sci. Forum 88-90, 27 ͑1992͒.
14 L. M. Di, G. F. Zhou, and H. Bakker, Phys. Rev. B 47, 4890
͑1993͒.
33 P. Duwez and F. Odell, J. Electrochem. Soc. 97, 299 ͑1950͒.
34
¨
H. Krainer, Arch Eisenhuettenwes. 21, 119 ͑1951͒.
35 M. D. Banus, T. B. Reed, and A. J. Strauss, Phys. Rev. B 5, 2775
͑1972͒.
15 G. F. Zhou, L. M. Di, and H. Bakker, J. Appl. Phys. 73, 1521
͑1993͒.
36 A. N. Cristensen, Acta Chem. Scand. A 32, 89 ͑1978͒.
37 W. C. Hamilton, Acta Crystallogr. 18, 502 ͑1965͒.
38 H. Zhang and E. H. Kisi, J. Phys. D 29, 1367 ͑1996͒.
39 P. Ehrich, Z. Anorg. Chem. 259, 1 ͑1949͒.
40 A. Brager, Acta Physicochem. CRSS 11, 617 ͑1939͒.
41 R. M. Haag and F. J. Shipko, J. Amer. Chem. Soc. 78, 5155
͑1956͒.
16 J. S. C. Jang and C. C. Koch, J. Mater. Res. 5, 498 ͑1990͒.
17 Y. S. Cho and C. C. Koch, J. Alloys Compds. 194, 287 ͑1993͒.
18 E. Hellstern, H. J. Feeht, Z. Fu, and W. L. Johnson, J. Appl. Phys.
65, 305 ͑1988͒.
19 L. M. Di, H. Bakker, and F. R. de Boer, Physica B 182, 91
͑1992͒.
42 R. J. Wasilewski and G. L. Kell, J. Inst. Met. 83, 94 ͑1954/1955͒.