J. Am. Ceram. Soc.

, 70 [6] 383-87 (1987)

The System Hf02-Ti02

JEAN PIERRE COUTURES and JULIETTE COUTURES
Centre de Recherche sur la Physique des Hautes TempCratures, CNRS, 45071 OrlCans Ccdex 2. France

The system HfOz-TiOz was studied in the 0 to 50 mol% TiOz solar furnace. It is clear that these results lead to a major discrep-
region using X-ray diffraction and thermal analysis. The ancy: the only reported intermediate compound, Hff i04(ortho-
monoclinic ( M ) S tetragonal ( T ) phase transition of HfOz was rhombic Phcn4), melts congruently around 2900°C after Ruh
found at 1750" +- 20°C. The definite compound HtTiOa melts et al., but according to Schevchenko et al. it shows an incongru-
incongruently at 1980" +- 1O"C, 53 mol% TiO,. A metatectic at ent melting behavior.
2300" f 20"C, 35 mol% TiO, was observed. The eutectoid de- Because of their very low coefficient of thermal expansion,' "'.'
composition of HfO,, v s ( T )+ HfO,, &W) + H f f i 0 4 , , , oc- there has been much interest in hafnia-titania ceramics having
curred at 1570" f 20°C and 22.5 mol% TiO,. The maximum a HfOz concentration >50 mol%. Therefore, according to the
solubility of TiOz in Hf02,,,(M) is 10 mol% at 1570" +- 20°C disagreement between the two liquidus lines of the published dia-
and in HfOz,,,(T) is 30 mol% at 1980" +- 10°C. On the grams, a reevaluation of the 0 to SO mol% TiO, region was under-
HfOz-rich side and in the 10 to 30 mol% TiO, range a second taken and the results are given below. As part of this work thc study
monoclinic phase M ' of HfOz(M) type was observed for sam- of HfO, polymorphism and particularly the temperature of thc
ples cooled after a melting or an annealing above 1600°C. The monoclinic ( M ) tetragonal ( T ) phase transition was also under-
phase relations of the complete phase diagram are given, using taken. Throughout this paper M , T , and C refer respectively to
the data of Schevchenko etal. for the 50% to 100% TiO, monoclinic, tetragonal, and cubic hafnia, and 0 to orthorhombic
region, which are based on thermal analysis techniques. HfTiO,.

I. Introduction 11. Experimental Procedure

T WO phase diagrams have been published for the Hf02-Ti02

system (Ruh et al., I Schevchenko et a / .') and are shown in
Fig. I . The first one (Fig. l(A)) was established by Ruh et al.'
The purity of the starting powdered oxides was as follows: Ti02,
99.9%;* HfOz (without Ti02), 99.9%.' The 0 to SO mol% TiO,
portion of the system was studied with samples at 5 mol% inter-
using for the melting point determinations a technique described by vals. The samples were obtained in air by comelting (melting time
Domagala ef al. and including room-temperature and high- 30 s) mixtures of the starting oxides with a 2-kW vertical axis solar
temperature phase analysis, differential thermal analysis, lattice furnace.' One sample set was quenched at around 1OOO"C-s-',
parameter determinations, and axial and bulk thermal expansion another set was cooled slowly (-l"C*s-'), and a third quenched
measurements. The second one (Fig. I ( B ) )was based on the data at 250°C * s-' . These cooling ratcs refer only to the liquid statc and
obtained with a thermal analysis technique using a horizontal axis the phase identification on these sample leads to information on the
shape of the liquidus line. The samples of the last set were ground
and annealed at 1300", 1600°, and 1700°C under oxygen
(Po? = 10' Pa) or argon (Po, = 0.6 Pa) and then quenched. These
treatments were carried out in order to confirm the observation of
Ruh et a/. I of a second monoclinic Hf0,-type phase in the 10 to
Received February 5, 1985; revised copy received February 10, 1986; approved 35 mol% TiOz and to study the effect of the oxygen pressure on the
November 26, 1986.
*Thann et Mulhouae, Thann, France. observed phases (i.e., reduction of TiO,) since TiO, is known
'Produits Chimiques Ugine Kuhlmann, Paris, France. to undergo reduction under mildly reducing conditions whereas

Fig. 1. Proposed HfO,-TiO:

phase diagrams: ( A ) Ruh et al.;
( B ) Schevchenko et al.

383
384 J o u r n a l of the American Ceramic Society-Coutures and Coutures Vol. 70, No. 6

Table I. Phases Observed in HfOz-TiOz Samples in the 0 to 50 mol% TiOz Range vs the Thermal Treatments*
Molten ramples annealed and quenched Sample, cooled from the melt
TiO, (mol%) 1300°C 1600°C 1700°C Slow cooling, I T . \ - ' Fact coolinp, I000"C.s-'
~

0 M M M M M
5 M M M M M
10 M + 0' M + 0' M +O' M + M' + 0' M + M ' + 0'
15 M + O M + M ' + O M + M ' + O M + M ' + O M + M ' + O
20 M + O M + M ' + O M + M ' + O M t M ' + O M + M ' + O
25 M + O M' + M ' + 0 M' +
M' + 0 M + M ' + O M + M ' + O
30 M + O M' + 0 M'+O M +MI + O +Ti+ M + M ' + O
35 M + O 0 M" +0 M+M'+O-tTi M + M ' + 0 + Ti'
40 M + O 0 0 M + M ' + 0 + Ti M + M ' + 0 + Ti
45 0 0 0 M + M ' + 0 + Ti M + M ' + 0 + Ti
50 0 0 0 M + M ' + 0 + Ti M S M ' + O +Ti
*M: monoclinic HfO, solid solution; M ': metastable monoclinic HfO, solid solution: 0 . orthorhomhic HITiO, solid rolution; Ti: TiO, rutile-type solid solution 'Traces.

I
i 200
M+M'
051 -
*-
I - i

1 300 "C 1600°C

0.50-

- * -
36 34 32 30 28 26 24
20
Fig. 2. X-ray powder diffraction pattern of a sample (composi-
tion 20 mol% TiO,) quenched after annealing at 1700°C in oxygen.

HfOz is stable under these conditions. Phase identification for all

B O ~ O O O B U O O O O i x o u o o o
samples was undertaken using room-temperature X-ray diffrac- v -
tion (Ni-filtered CuKa radiation; goniometer scanning speed 1" M+O 0 E"?'y7F?'E
Lo'
20 min-I). The phases present from room temperature up to
2000°C were determined using a high-temperature diffraction 30 LO 50
, Y
0 O,
30 LO 50
I ! , 30 I

40 50
camera with a ribbon of platinum up to 1500°C or rhenium as the Ti02 mole %
heating element at higher temperatures. In this case experiments
were carried out in a He-2% H2 atmosphere for the 1500" to 2000°C Fig. 3. Evolution of the cell parameters of HfTiO,,,(O) vs the tem-
range. The sample was finely ground and deposited on the heating perature of annealing for samples in the 25 to SO mol% TiO, range.
element. Below 1500°C runs were performed under air and the
temperature was measured with an ultrafine thermocouple spot-
welded on the heating element.' Above 1500°C an optical micro- our study, the melting times were short (20 s), leading to weight
pyrometer (A = 0.65 pm) was used. In the 1100" to 1500°C range losses of <0.5%, except for the slowly cooled samples where an
the temperature was simultaneously measured with the thermo- average compositional change not greater than 3 mol% TiOz
couple and the micropyrometer; agreement between the two was occurred. For the firings no weight change was observed. No
obtained for an emissivity of 0.7. This emissivity was used for the chemical analyses were performed and the chemical shifts were
temperature measurements over 1500°C. The liquidus and solidus estimated by measuring the weight losses during each thermal
lines were drawn from the results obtained using a thermal analysis treatment; as a consequence the nominal compositions were not
device,' which was combined with a 2-kW horizontal axis solar corrected.
furnace. The temperature was measured with an optical pyrometer
under blackbody conditions' (A pyrometer 0.8 1 pm: practical 111. Results and Discussion
ernissivity 0.99; calibration points: AI2O3 mp 2053"C, YzOi mp
(1) Phase Relations
2440°C). The temperature (68 IPTS) for the liquidus line is 2 10°C
up to 2300°C and 220°C over 2300°C. (A) P h a s e Identification at Room Temperature: Room-
In such studies and particularly at high temperatures the prefer- temperature X-ray diffraction patterns were obtained for all
ential vaporization of TiO, leads to a possible shift of the sample samples. The results, on samples melted or fired at 1600" and
composition: this point was not outlined in the previous works. In 1700°C. and shown in Table I , support the presence of a second
June 1987 The System Hf0,-TiO, 385

Table 11. Evolution of Cell Parameters of H f f i O , , , ( O ) on Samples Annealed at 1600" and 1700°C and Quenched vs P,,*
1600°C 1700°C
10' Pa 0 6 Pa 10' Pa 0 6 Pd
Ti0, U h C U h c (1 h < 11 h <
(mol%j
~
fnmi
~
Inm)
~~~~
Inm) O/(HT + TI)
~~
Inm) fnm) Inmi Inm) (nm) fnmi
~ ~
O/IHf
~
+ TI)
~~~~~~~~
(nml (nm) (nml

100 1.98" 1.966

50 0.4756 0.5563 0.5045 1.993 0.4759 0.5564 0.5048 0.4756 0.5563 0.5047 1.987 0.476, 0.5573 0.5058
45 0.4776 0.5561 0.5061 1.99, 0.477, 0.5568 0.5062 0.4768 0.5565 0.5062 1.992 0.4778 0.5568 0.508,)
40 0.4792 0 . 5 5 8 ~ 0.5070 1.997 0.4794 0.5580 0.507, 0.4785 0.5584 0.5074 1.994 0.479s 0.559,, 0.508,)
0 [Link] 2.000

Table 111. Temperature of the Hf02(M) Z HfOz(T) Phase

Transition and Cell Parameters of the Tetragonal Phase
Authors T ("C) a (nmj b (nrnj cia
Aldebert et $ . ' I 1765 0.5149 0.5294 1.028
Stacy et al. 1763 0.514s 0.5297 1.029
Pa This work 1780 0.5153 0.5296 1.028

100

48
97 poses by the following reaction:
HfOz,,(M') + Hf02,,,(M) + HtTiO,. J O )
X-ray examination at room temperature of slowly cooled samples
shows that traces of a rutile phase (TiOz) appear at 30 mol% TiOz.
The concentration of this rutile phase increases in the samples
richer in TiOz. This indicates that HffiO, does not melt congm-
I145 ently and also that the maximum solubility of T i 0 2 in tetragonal
Hf02.,, is around 30 mol%.
For the 25 to 30 mol% TiO, region Fig. 3 shows ( I ) the evo-
lution of the room-temperature cell parameters for the ortho-
'nm3 rhombic H f f i 0 4 solid solution and (2) the nature of the observed
phases for samples annealed at 1300", 1600", and 1700°C. After
these treatments no new phases were observed for samples fired at
1.140 1600" and 1700"C, under either oxygen or argon. Neverthcless,
micro TGA examinations show some loss of oxygen in the speci-
L p' mens with high TiOz content after the annealing under argon

Y
//---- (Po, = 0.6 Pa) and therefore slight increase of the cell parameters
is observed for specimens fired in argon (Table 11).
( B ) Phase Identification as a Function of Temperature: The
martensitic monoclinic Z tetragonal (M Z T ) transformation
I I temperature and melting point of HfOl were determined by Ruh
1000 1500 2000 2500 ("C 1 et al. They obtained a value of 1750°C for the transformation with
a smaller hysteresis than Zr02 and a value of 2900°C for the
Fig. 4. Lattice parameters vs T for HfOz.
melting point. This M Z T phase transition was reexamined and
the evolution of the cell parameters is shown in Fig. 4. The transi-
tion temperature, 1750" + 20"C, and the cell parameters
(Table 111) measured in this work agree very well with previous
monoclinic Hf0,-type phase called M ' as suggested by Ruh et a / . ' measurements (Stacy et al., In Aldebert et al. 'I). This correlation
Figure 2 clearly shows some peaks due to M ' observed on a sample provides a justification for the 0.7 values of emissivity used in thc
of 20 mol% TiOz heated at 1700°C in oxygen and quenched; the temperature measurements. Figure 4 shows clearly that the b
cell parameters of the two monoclinic phases are respectively parameter of the HfOz(M) monoclinic cell is independent of the
a = 0.528] nm, b = 0.507y nm, c = 0.505, nm, p = 98.63", temperature (i.e., within the experimental error of the parameter
V = 0.134,) nm'; u = 0 . 5 2 7 1 n m , b = 0.513, n m , calculation +-0.0005 nm). The volume contraction due to the
ic = 0.5078 nm, p = 99.22", and V = 0.135, nm3. Annealing HfOz(M) Z Hf02(T) transition was found to be 2.7%. The phases
this specimen at 1300°C shows the disappearance of one of the two present in the 0 to 50 mol% TiOz were determined from room
monoclinic HfOz-type phases (i.e., the phase with the lower pa- temperature up to 2000°C by high-temperature X-ray diffraction
rameter) while the intensity of the reflection peaks of Hf02,,,,(Mj using the samples annealed at 1300°C for 24 h. During these ex-
and HffiO,, JO) increases. The decomposition occurs slowly and periments and particularly for T > 1500°C, in order to avoid com-
takes 8 h at 1300"C, but at 1250°C the two monoclinic phases are position shifts due to the preferential TiOz vaporization, the time
still present after a 24-h heating. For the samples cooled after a was minimized by recording only the important portions of the
melting or an annealing above 1600°C a similar behavior is ob- diffraction patterns. The results are given in Fig. 5 and outlined
served in the 10 to 30 mol% TiO, range. In any case the annealing below with compositions at +2 mol%:
at 1300°C does not modify the parameters of the monoclinic The limits of the HfO,, &Vf) solid solution at 1300", 1600", and
Hf0,-type phase with the highest volume of the cell. The X-ray 1700°C are respectively 7, 9, and 5 mol% TiOz.
diffraction results indicate that V , , < V,; therefore, as rT,44 < At the same temperatures the limits of the Hffi04,.s,5(0)solid
rHf4+the concentration of TiOz is higher in the M ' than in the M solution on the HfOz-rich side are respectively 42, 33, and
phase. According to our results, at 1300°C the M ' phase decom- 37 mol% TiOz.
386 Journal of the American Ceramic Society-Coutures and Coutures Vol. 70, No. 6

a,b,c ( n m ) Hf12T'OSOL Vnm3

O T
M+O
056; -&--A
/-
I
0 M+T
O+T 0 55- 0 139
2WLl 0 0 u
0%- -0138

053 - - 0 137
- 0 136

-
1 - - * L

- - -- - _
I 0.51 9 -
c - - 0 135
I \ (C I
I O \ O
I I 050-

100 =' 010 o

Fig. 5. Phase identification by high-temperature X-ray vs the

composition. 500 lo00 1500
T ("C)

Fig. 6. Evolution vs the temperature of Hffi04,,, lattice parameters for

a 40 mol% TiO, sample.
Table IV. Liquidus Temperature and Thermal Arrest vs
the TiOz Content for Compositions up to 50 moi% TiO,
in the HfO,-TiO, Svstem
TiO, (mol%) Liquidus T ("C) Thermal arrest ("C)
0 2850 2600 (Hf02(C) Z Hf02(T))
10 2800 2520 (solidus)
20 2700 2310*
25 2600 2300*
30 2480 2300*
35 2300 1990t
40 2250 1980'
45 2200 1980'
50 2100 1980'
*Metatectic line. 'Peritectic line

The phase boundary of the two-phase region HfOr..,$(M) +

HfTiO,.,,) g i v i n g e i t h e r HfOz.,,(M) + Hf02,.,(T) or
HfO,,,,(T) + Hffi04.,, is 1570" 5 20°C.
The composition of the eutectoid point HfO,,,,(T) S
Hf02.,,(M) + HfliO,,,, is 18 mol% T i 0 2 .
At 1570°C the maximum solubility of TiOz in Hf02.,,(M) is
10 mol% T i 0 2 and the maximum solubility of HfOz in HffiO,, s , ~
corresponds to 32 mol% T i 0 2 .
The high-temperature X-ray diffraction technique allows one to
measure precisely the evolution of cell parameters vs the tem-
perature. Accurate measurements of the microscopic thermal
expansion of HffiO, could be carried out and some of these
measurements are given in Ref. 12. Figure 6 shows the results
obtained for a two-phase sample (HfOZ.,,(M) + H f f i 0 4 . < , )with
an average T i 0 2 content of 40 mol%. The data agree with the Fig. 7. Tentative Hf0,-TiO, phase diagram.
results of Ruh et ul. ' Carlson rt al. ' and Skaggs6
(2) Phase Diagram
Cooling curves obtained with the thermal analysis apparatus the high-temperature X-ray results, yield the partial subliquidus
show and confirm the X-ray results: no compounds with a congru- phase diagram given in Fig. 7. Our data for the liquidus line of the
ent melting behavior were found in the 0 to 50 mol% T i 0 2 range. Hf02-Hffi04 region agree well with the results of Schevchenko
The liquidus temperature and the thermal arrest for the samples et al. ; therefore, we have included as dotted lines their results in
studied are given in Table IV. These data, along with the phase the HtTi04-TiOz part in order to present a complete phase dia-
identification obtained by room-temperature X-ray diffraction and gram. Above 2300°C the diagram was constructed to be consistent
June 1987 The System HfO,-TiO, 387

Table V. Invariant Points and Their Assignments in the A possible explanation could be found in the fact that these
HfOz-Ti02System authors believe that their melting point measurements should be
T (%) TiO, (mol%) Nature considered to be qualitative rather than quantitative. In the sub-
2850 0 HfO, melting point solidus region comparison could only be made with the work of
2600 0 Hf02(C) S HfOz(T) transition Ruh et al. As they stated, for solid samples in the 10 to 30 mol%
2300 10 Equilibrium HfO,, ,JC)- TiO, range heated over 1600°C and quenched, as with molten
Hf02..,(T)-Liq specimens cooled either rapidly or slowly, a metastable Hf0,-type
12 Hf02,,,(C) Z HfO2.,,(T) + Liq monoclinic phase M ’ was obtained which disappeared after the
35 Metatectic point appropriate firing below 1570” ? 20°C. This observed phase is
1980 30 Max solub of TiOz in due to the greater solubility of TiO, in the unquenchable tctragonal
Hf02,,(T); equil form of Hf02 than in the monoclinic form (=30 mol%, 10 mol%).
HfO2.,,(T)-HffiO4,,,(O) Our eutectoid temperature, 1570” -C 20”C, agrees rather well with
-Liq the data of Ruh et al. (1600°C) but we disagree for the eutectoid
50 HffiO, noncongruent melting
point composition (our work, 22.5 mol% Ti0,; Ruh et al., =I0 mol%
53 Peritectic point’ TiO,). Concerning the phase boundaries it appears essentially that
1750 0 HfO,(TI S Hf02(M) transition the range of existence of HffiO,,, is more narrow than in the
1740 60 Max-solub of TiO, in previous determination. As a part of this work the HfO,(M) 2
Hff104,..2 Hf02(T) transition temperature (1750” -C 20°C) was measured and
75 Hff i04,ss-TiOZ, eutectic’ the obtained results agree fairly well with the published data.
92 Max solub of HfOt in TiO2.,
1570 10 Max solub of TiO, in
HfOz,..(M)
22.5 Eutectoid Hf02,..(T) 3 Acknowledgments: This work was carried out on behalf of the French Cana-
HfO,.,,(M) + Hfli04,,,(0) dian OCTET Group during the authors’ stay at the Super-Refractory Laboratory. G
34 Max solub of HfOz in Benezech and R. Renard of this Laboratory are gratefully acknowledged. Thanks are
HffiO,.,(O) due to the OCTET members and to Dr. S . R. Skaggs of the Los AIamos National
Laboratory for their encouragement and fruitful discussions.

with the phase rule. Only the temperature and the composition of References
the metatectic due to the HfO>(C)S Hf02(T) transition are ‘R. Ruh, G. W. Hollenberg, E. G . Charles, and V. A. Patel, “Phase Relations and
Thermal Expansion in the Hf0,-TiO, System,” J . Am. Cerum. Soc., 59 [11-12]
known (2300” t 20°C; 35 mol% TiO,). 495-99 (1976).
The phase boundaries HfOz,,,(T)/Hf02.J T ) + HfOz,TT(C), *A. V. Schevchenko, L. M. Lopato, A. I. Stegny, I. M. Maister, V. S. Dverniakov,
and V. V. Passitchny, “Liquidus of the HfO,-TiO, and Hf0,-Eu,O, System,” Dokt.
HfO,. T s ( T )+ H f 0 2 , T v ( C ) / H f O ~ . . , . r ( Cand
), Hf02..,(C)I Akpd. Nuuk SSSR, Ser. A , 87, 58-85 (1979).
Hf02,,,(C) + Liq could not be ascertained first because of R. F. Domagala and E. Heckenbach, “High Temperature Furnace for Melting
the composition shifts due to TiOz vaporization when high- Point Determination,” Rev. Sci. Instrtrm., 35 1121 1663-64 (1964).
4A. Harari, J. P. Bocquet, M. Huber, and R . Collongues, “Crystal Structure of the
temperature X-ray diffraction experiments are carried out above Mixed Oxide HtTiO,,”C. R . Hebd. SeancesAcad. Sci., Ser. C . 267, 1316-18 (1968).
2300°C and second because no clear thermal arrest except the ’ G . A. Carlson, J. L. Anderson, R. A. Briesmeister, S. R. Skaggs, and R. Ruh,
“Coefficient of Thermal Expansion and Dynamic Response to Pulsed Energy Deposi-
solidus transition at 2520°C for the 10 mol% T i 0 2 composition is tion in Hf0,-TiO, Compositions,” J. Am. Cerum. Soc.. 60 [11-12] 508-10 (1977).
observed on the cooling curve. The different invariant points and 6(a) F. H. Simpson, “High Temperature Structural Ceramics,’’ M a w . Des. E n g . ,
their assignments are given in Table V. 52 [2] 1 6 1 8 (1960). (b) S . R . Skaggs, “Zero and Low Coefficient of Thermal
Expansion of Polycrystalline Oxides,” Rev. Int. Hautes Temp. Refracr., 16, 157-67
(1979).
’3. P. Coutures, R. Berjoan, G . Benezecb, B. Granier, R. Renard, and M. Foex,
IV. Conclusion “Uses of Laboratory Scale Furnace for High Temperature Studies of Physical and
Chemical Properties of Refractory Oxides,” Rev. Int. Hautes Temp. Refract , 15,
A tentative phase diagram for the system HfOz-TiOz has been 103-14 (1978).
8 J . Coutures and J . P. Coutures, “Study of the Phase Transition of LnAIO, (in
determined using in the 0 to 50 mol% TiO, region the data ob- Lanthanidic Element) Perovskites by High-Temperature X-ray Diffraction ,” J . Solid
tained in our study and in the 50 to 100 mol% TiOz region results Stufe Chem., 5 2 , 95-100 (1984).
of Schevchenko et al. ’ This work clearly shows that HffiO, does 9F. Cabannes, J. Simonato, M. Foex, and J. P. Coutures, “Use of Eutectic, Melting
Points as Secondary Temperature Standard Above 2327 K,” High Temp. - High
not have a congrucnt melting behavior but decomposes at 1980” -C Pressures, 4, 589-96 (1972).
20°C to HfOz(T) + Liq. As a consequence no eutectic exists be- I’D W. Stacy, J . K. Johnstone, and D. R. Wilder, “Axial Thermal Expansion of
HfO,,” J . Am. Ceram. Soc., 55 [9] 482-83 (1972).
tween HfOz and Hfl30,. In the 0 to 50 mol% TiOz the proposed “P. Aldebert, J. M. Badie, J . P Traverse, J . L. Buevoz, andG. Roult, “Application
liquidus line agrees with the work of Schevchenko et al. but dis- of a High Temperature Neutron Diffraction Apparatus to the Study of Refractory
Oxides,’’ Rev. Int. Hautes Temp. Refract., 12, 307- 19 (1975).
agrees with the estimates of Ruh et a / . particularly in the 30 to ‘*L’OctetFranco-Canadien, “Chdracterisation of Some Materials in the Hf0,-TiO,
50 mol% TiO, range. System,” High Temp.-High Pressures, 13, 97-103 (1981). U