Formation of zirconium diboride (ZrB2) by room temperature mechanochemical reaction between ZrO2, B2O3 and Mg

Article abstract of DOI:10.1016/j.jallcom.2005.07.083

A mixture of magnesium, boric oxide and zirconium dioxide were mechanically milled under argon for up to 15 h in a laboratory scale ball mill. X-ray diffraction showed that there was an increasing conversion of ZrO₂ to ZrB₂ with milling time with >98% reaction after 15 h. Differential thermal analysis revealed there were multiple, overlapping reactions all of which seemed to be formation of ZrB₂. The energy evolved decreased with milling time and the sample after 15 h milling showed no thermal reaction. After milling, separation of the ZrB₂ from the coproduct MgO was easily achieved by a mild acid leaching leaving essentially pure ZrB₂ with a crystallite size of ～75 nm.

Full text of DOI:10.1016/j.jallcom.2005.07.083

Journal of Alloys and Compounds 420 (2006) 225–228
Formation of zirconium diboride (ZrB2) by room temperature
mechanochemical reaction between ZrO2, B2O3 and Mg
N. Setoudeh1, N.J. Welham∗
Extractive Metallurgy, Murdoch University, Perth, WA 6150, Australia
Received 25 March 2005; accepted 25 July 2005
Available online 15 December 2005
Abstract
A mixture of magnesium, boric oxide and zirconium dioxide were mechanically milled under argon for up to 15 h in a laboratory scale ball mill.
X-ray diffraction showed that there was an increasing conversion of ZrO2 to ZrB2 with milling time with >98% reaction after 15 h. Differential
thermal analysis revealed there were multiple, overlapping reactions all of which seemed to be formation of ZrB2. The energy evolved decreased
with milling time and the sample after 15 h milling showed no thermal reaction. After milling, separation of the ZrB2 from the coproduct MgO
was easily achieved by a mild acid leaching leaving essentially pure ZrB2 with a crystallite size of ∼75 nm.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Zirconium diboride; ZrB2; Zirconia; ZrO2; Mechanochemistry
1. Introduction
by SHS [8]. However, the use of elemental Zr and B is much
more expensive than using the oxides due to the high cost of
extracting metals from the minerals. A similar argument regard-
ing costs can be made for recent processes which use complex
polymeric metal alkoxides or air and moisture sensitive materi-
als, suchasborazine, astheprecursors[9,10]. Thesepreparations
also require pyrolysis before the required phases are formed.
Recently, there have been a number of studies on ambient-
temperature reduction reactions which occur during extended
ball milling to form ceramic products [11–13]. Welham synthe-
sised <200 nm, single crystal particles of titanium diboride after
15 h of milling of a mixture of titanium dioxide, boron oxide
and pure magnesium [14]. The use of magnesium as the reac-
tant instead of aluminium allowed acid leaching of the unwanted
oxide product leaving behind a pure boride phase. As there was
no product at 10 h but complete reaction at 15 h the reaction was
thought to have required an incubation time before becoming
self-propagating. Theparticlesformedwereclearlyfacetedcrys-
tals which implies a high temperature must have been present in
the mill during reaction.
The diborides of titanium (TiB2) and zirconium (ZrB2) are
materials suitable for high performance applications. They have
high melting points (3200 and 3250 ◦C) and high hardness val-
ues (30.0 and 36.0 GPa), respectively. These materials also have
high resistance to oxidation at elevated temperatures, an excel-
lence chemical resistance to HCl and HF and are stable in
metal melts (Al, Cu, Mg, Zn, Cd, Fe, Pb), cryolite and non-
basic slags [1–3]. Composites using zirconium diboride with
alumina have mechanical strengths similar to that of a titanium
diboride–aluminacomposite[4,5]butareresistanttohighertem-
peratures.
The self-propagating-high temperature synthesis (SHS) pro-
cess is one advanced method which has been used extensively
for preparing refractory materials such as carbides, silicides,
nitrides and various composite materials [6,7]. Mishra et al.
synthesised a zirconium diboride–alumina composite by SHS
process of a milled and pelletised mixture of zirconium oxide,
aluminium and boron oxide [3]. Radev and Klissurski used pure
elemental zirconium and boron to synthesis zirconium diboride
This article demonstrates the formation of zirconium diboride
by ball milling a mixture of oxides and pure magnesium with
the stoichiometry of reaction (1):
∗
Corresponding author. Tel.: +61 8 9360 2831; fax: +61 8 9360 6343.
E-mail address: nicholas.welham@murdoch.edu.au (N.J. Welham).
Present address: Mechanical Engineering, Yasouj University, Yasouj 75914,
Iran.
ZrO2 + B2O3 + 5Mg = ZrB2 + 5MgO
1
ꢀH = −957 kJ at 25 ◦C
(1)
0925-8388/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2005.07.083

226
N. Setoudeh, N.J. Welham / Journal of Alloys and Compounds 420 (2006) 225–228
The highly exothermic nature of the reaction could be used
to monitor the progress of the reaction by measuring the vial
temperature as a function of time. The temperature would be
expected to change regardless of reaction, primarily due to fric-
tion and impact energy transfer due to the motion of the balls.
By this mechanism, the temperature would be expected to rise
until heat loss by conduction (along the drive shaft), convection
(natural and due to mill rotation) and radiation were equal to the
physical energy generated. This would be expected to be a fairly
constant rise.
For a self-propagating reaction, a single large temperature
spike would be expected after the required incubation time as
all exothermic energy would be evolved within a few seconds.
For the 7.0 g of reactant mixture used a total of 21.7 kJ of energy
would be release by the complete reaction. If this energy is used
to adiabatically heat the product and milling vial (∼2.5 kg) from
25 ◦C, the equilibrium temperature would be 44.3 ◦C. However,
if the reaction proceeds at a constant rate, then there would be
no spike, simply more rapid attainment of an equilibrium tem-
perature.
Fig. 1. Milling vial temperature as a function of time.
slightly surprising as there is no apparent endotherm at 650 ◦C
where any residual Mg would melt. The 2 h milled sample also
shows an exotherm starting around 400 ◦C but it was somewhat
wider than that after 8 h milling, this is a complex exotherm indi-
cating that the conversion of ZrO2 was not a simple process. The
reaction was clearly slowing above about 520 ◦C but reignited
around 650 ◦C, this was probably due to an increase in interfacial
reaction area due to the melting of Mg. A final exotherm is also
evident above ∼750 ◦C. XRD of the residue showed increased
conversion of ZrO2 compared with the as-milled powder, but
less than the 8 h as-milled powder. In all cases, the residue after
heating contained both monoclinic and tetragonal ZrO2, with
tetragonal becoming increasingly dominant with milling time.
It has been shown previously [14,15] that a self-propagating
reaction results in a narrow and very intense exothermic peak
during DTA due to the near instantaneous combustion of the
entire sample. In Fig. 2, the presence of multiple, broad, rela-
tively small exotherms is indicative of a system which reacted
2. Experimental
Mixtures of raw materials were prepared in accordance with the stoichiom-
etry given by reaction 1 (39.2 wt.% ZrO2, 22.1 wt.% B2O3 and 38.7 wt.% Mg).
Seven grams was loaded into the mill with five 25.4 mm diameter stainless steel
balls giving a ball to powder mass ratio of 43:1. The vial was evacuated to
10−2 Pa then backfilled with argon to prevent oxidation of the magnesium and
placed onto the mill drive where it was run at 165 rpm for up to 15 h. An infrared
thermometer was used to monitor the temperature of the milling vial.
After milling, samples were heated to 1000 ◦C in an alumina crucible at
20 ◦C/min under a flowing argon (100 mL/min) atmosphere using a differen-
tial thermal analyser (DTA). Simultaneous thermogravimetry revealed no mass
losses were associated with the experiments.
The magnesium oxide and any residual magnesium and/or boron oxide were
removed by leaching the as-milled powder in 10% hydrochloric acid for 1 h. The
solution was decanted after leaching and deionized water was added to wash the
solids and the residue dried in air. All of the products were analysed by XRD
using Co K␣ radiation over a range of 20–80◦ 2θ, using a step size of 0.1◦ and
3 s counting time per step.
3. Results and discussion
The temperature of the vial with increasing milling time is
shown in Fig. 1. Apart from the expected underlying increase the
major feature is a rise at 115 min, clearly, there is an exothermic
reaction occurring, although the temperature rise of just over
3 ◦C implies the reaction was relatively minor. Samples were
milled for 115 and 130 min and the products analysed by XRD.
However, no discernible differences between these traces were
observed indicating the reaction was relatively minor. After this
peak there are no more clearly defined thermal events which
imply that the majority of the reaction was gradual rather than
self-propagating as in the case of TiB2.
The DTA data presented in Fig. 2 shows that there were no
thermal events in the sample milled for 15 h, clearly, the reac-
tion was at its maximum extent after milling. The sample milled
for 8 h showed an exotherm starting at around 400 ◦C, XRD of
the residue showed a greater conversion of ZrO2 than in the
as-milled powder, but the reaction was still incomplete. This is
Fig. 2. Differential thermal analysis of the as-milled powders. Data has been
normalised to the same mass and then offset for clarity.

N. Setoudeh, N.J. Welham / Journal of Alloys and Compounds 420 (2006) 225–228
227
Fig. 3. XRD traces of powders milled for up to 15 h. Also shown are traces for
15 h milled samples heated to 1200 ◦C and as-milled powder leached for 1 h in
0.1 M HCl.
Fig. 4. Fraction of zirconium present as ZrB2.
gradually and therefore would not be expected to undergo self-
propagation since the rate of release of energy is too low to
heat any nearby reactants beyond the temperature required for
ignition.
and this data is presented in Fig. 4. This graph reinforces the
gradual nature of the reaction with conversion of ZrO2 to ZrB2
being a direct function of milling time.
The non-zero intercept of the linear fit indicates that in the
early stages there is a disproportionate rate of formation of ZrB2.
This not entirely unexpected since in the early stages particles
of ZrO2, B2O3 and Mg are more likely to react as the parti-
cle surfaces will be fresh. After some time the particle surfaces
will be composed primarily of product thereby decreasing the
probability of reaction.
The gradual nature of this reaction is clearly different to the
analogous formation of TiB2 from TiO2 reported previously
[14]. The milling equipment and conditions were identical yet
TiB2 formed by a high temperature process while ZrB2 formed
gradually.
Thermodynamically, reaction (1) has a free energy of
−3.0 kJ/g and an enthalpy of −3.0 kJ/g, the same reaction for
TiO2 has a free energy of −3.8 kJ/g and enthalpy of −3.9 kJ/g
all at 25 ◦C. Thus, for the same mass of starting mixture, TiO2
is more favourable and produces more energy than ZrO2, yet
the rate determining step is apparently different. If it is assumed
that the energy generated by reaction (1) is used to heat the reac-
tants, then a temperature of >1000 ◦C is obtained for both TiO2
and ZrO2 systems. For TiO2 this is above the ignition temper-
ature of 1 h milled powder [14] and above the exotherms for
ZrO2 shown in Fig. 2. Consequently, once initiated, the reac-
tion could be expected to be self-propagating, leading to a near
instantaneous reaction within the mill. Therefore, it seems that
for the TiO2 system the ignition temperature for the mixture
was attained but not in the ZrO2 system. However, it could be
argued that there was a minor stage of self-propagation where
the mill heated up, as shown in Fig. 1, however, the extent
of this reaction remained low and the remaining ZrO2 reacted
gradually.
XRD traces of as-milled powders are shown in Fig. 3. After
2 h of milling it is clear that the major phases present are the
starting tetragonal zirconia (t-ZrO2) and elemental magnesium.
The absence of peaks for B2O3 is not unexpected as the starting
B2O3 showed no discernible peaks, just a broad peak around 30◦
indicating a glassy phase. There was a small peak at 50.3◦ due
to MgO, however, this could have formed by oxidation of the
milled magnesium on exposure to atmosphere after milling. The
presence of peaks for ZrB2 at 29.6◦, 38.6◦ and 48.8◦ indicate
that some reaction had occurred.
Extending the milling time to 8 h showed the reaction was
more extensive with peaks for the starting phases weakening
as those for ZrB2 and MgO intensified. After milling for 15 h
the reaction was apparently complete with only peaks for MgO
and ZrB2 present. XRD of the sample heated to 1200 ◦C, Fig. 3,
showed narrowed and intensified peaks for the product phases.
Also present were a weak peak at 33◦ due to t-ZrO2 and a slightly
stronger peak at 35.3◦ which can be attributed to monoclinic
m-ZrO2. The latter polymorph of ZrO2 is formed by thermal
transformation of the tetragonal form, this transformation may
have been enhanced by the smaller crystallites formed during
milling [16–20]. The presence of ZrO2 indicates that the reac-
tion to form ZrB2 was incomplete. The absence of a melting
endotherm for Mg in the 15 h DTA trace in Fig. 2 implies there
was insufficient Mg to complete the reduction of ZrO2. In a prac-
tical application of this process excess Mg and B2O3 would be
used to account for this.
The absence of any unattributed peaks in the XRD traces
would imply that the reaction occurred by reaction (1) without
any formation of intermediate phases.
Profile fitting using Powder Cell [21] was carried out for the
threemillingtimesandtherelativeabundancesofthefourphases
(ZrO2, Mg, ZrB2 and MgO) determined. From the data for zir-
conium phases the fraction of Zr present as ZrB2 was estimated
Dissolution of the residual Mg and product MgO was readily
achieved using 0.1 M HCl, the XRD trace of the leached product
after 15 h milling is shown in Fig. 3. Clearly, the major phase
present in the solid residue is ZrB2, although the small peak at

228
N. Setoudeh, N.J. Welham / Journal of Alloys and Compounds 420 (2006) 225–228
33.3◦ is due to unreacted ZrO2. Determination of the unit cell
size of the ZrB2 formed was made on the peaks up to 2θ = 80◦
using UnitCell [22], the obtained values, with error in parenthe-
ses, were a = 0.31655(1) nm and c = 0.35265(1) nm. These are
slightly smaller than the published values of 0.316870(8) and
0.353002(10) nm [23] which are used in the Powder Diffraction
File. Crystallite size determination using the Scherrer equation
indicated a size of ∼75 nm.
Acknowledgement
The authors would like to thank the Iranian Government for
providing funding for NS to pursue this research.
References
[1] P. Schwarzkopf, R. Kieffer, Refractory Hard Metals: Borides, Carbides,
Nitrides and Silicides, Macmillan, New York, 1953.
[2] L.S. Sigl, K.A. Schwetz, in: D. Emin (Ed.), Boron Rich Solids, AIP
Conference Proceedings, vol. 231, American Institute of Physics, New
York, 1991, pp. 468–472.
[3] M.M. Opeka, I.G. Talmy, J.A. Zaykoski, J. Mater. Sci. 39 (2004)
5887–5904.
[4] S.K. Mishra, S.K. Das, P. Ramachandrarao, D. Yu. Belov, S. Mamyan,
Metall. Mater. Trans. A 34 (2003) 1979–1983.
Inany process for the formation ofZrB2 itwouldbenecessary
to ensure that all of the ZrO2 is consumed by reaction as ZrO2
is difficult to remove by dissolution. This could be achieved by
using a stoichiometric excess of both Mg and B2O3 in any pro-
cess. ApreviousattempttomakeMgB2 bymagnesiumreduction
of B2O3 was unsuccessful with no reaction evident even after
heating a 50 h milled sample to 1200 ◦C so any excess of these
phases would remain once the ZrO2 had been consumed and be
dissolved along with the MgO product.
[5] J. Liu, P.D. Ownby, J. Am. Ceram. Soc. 74 (1991) 241.
[6] D.V. Shtansky, E.A. Levashov, A.N. Sheveiko, J.J. Moore, J. Mater.
Synth. Proc. 6 (1998) 61–72.
The mill used in these investigations is known to be extremely
inefficient and a more efficient device would be expected to lead
to a substantially shorter milling time. This has recently been
demonstrated [24] with milling times of around 1 h required to
achieve the same result as between 10 and 15 h milling in the
mill used here [14]. Thus, it seems probable that formation of
ZrB2 can be achieved more rapidly than in this paper. It may
also be that a more rapid induction of reaction may lead to the
self-propagation of the formation of ZrB2 observed in the TiB2
system.
[7] A.G. Merzhanov, J. Mater. Proc. Technol. 56 (1996) 222–241.
[8] D.D. Radev, D. Klissurski, J. Mater. Synth. Proc. 9 (2001) 131–136.
[9] S.T. Schwab, C.A. Stewart, K.W. Dudeck, S.M. Kozmina, J.D. Katz,
B. Bartram, E.J. Wuchina, W.J. Kroenke, G. Courtin, J. Mater. Sci. 39
(2004) 6051–6055.
[10] K. Forsthoefel, L.G. Sneddon, J. Mater. Sci. 39 (2004) 6043–6049.
[11] P.E. Willis, N.J. Welham, A. Kerr, J. Eur. Ceram. Soc. 18 (1998)
701–708.
[12] A. Kerr, N.J. Welham, P.E. Willis, Nanostruct. Mater. 11 (1999)
233–239.
[13] N.J. Welham, A. Kerr, P.E. Willis, J. Am. Ceram. Soc. 82 (1999)
2332–2336.
[14] N.J. Welham, J. Am. Ceram. Soc. 83 (2000) 1290–1292.
[15] N.J. Welham, Mater. Sci. Eng. A 255 (1998) 81–89.
[16] S. Begin-Colin, T. Girot, A. Mocellin, G. LeCaer, Nanostruct. Mater. 12
(1999) 195–198.
[17] R. Ren, Z. Yang, L.L. Shaw, J. Mater. Sci. 35 (2000) 6015–6025.
[18] T. Girot, X. Devaux, S. Begin-Colin, G. LeCaer, A. Mocellin, Phil. Mag.
A 81 (2001) 489–499.
[19] J. Hu, H. Qin, Z. Sui, H. Lu, Mater. Lett. 53 (2002) 421.
[20] S. Begin-Colin, G. LeCaer, A. Mocellin, A. Zandona, Phil. Mag. Lett.
69 (1994) 1–7.
[21] W. Kraus, G. Nolze, J. Appl. Cryst. 29 (1996) 301–303.
[22] T.J.B. Holland, S.A.T. Redfern, Miner. Mag. 61 (1997) 65–77.
[23] R. Keissling, Acta Chim. Scand. 3 (1949) 90–91.
[24] R. Ricceri, P. Matteazzi, Mater. Sci. Eng. A 379 (2004) 341–346.
4. Conclusions
Zirconium diboride (ZrB2) has been synthesised by a low
temperature solid-state reaction between zirconium dioxide,
boron oxide and elemental magnesium. The reaction was found
to occur during ball milling in a tumbling mill with a gradual
conversion from ZrO2 to ZrB2 during milling. The conversion
was almost complete after 15 h milling and may have been hin-
dered by a slight sub-stoichiometry in Mg. The concomitantly
formed MgO was selectively separated by a simple acid leaching
step leaving behind an essentially pure ZrB2 powder.