242110-3
Roucka et al.
Appl. Phys. Lett. 89, 242110 ͑2006͒
TABLE I. Lattice parameters of Zr1−xHfxB2 films obtained from HR-XRD
analysis ͑bold font͒. The relaxed lattice parameters ͑a0 and c0͒ and calcu-
lated strains ͑ and ͒ are obtained from a compositional interpolation of
a
c
the elastic ratio ͑x͒ and bulk c/a ratio ͑x͒.
x
0
15
30
60
100
a
3.187
+0.52
3.521
−0.30
−0.584
1.114
3.170
3.531
3.183
+0.54
3.513
−0.32
−0.580
1.113
3.166
3.524
3.176
+0.50
3.505
−0.28
−0.574
1.112
3.160
3.515
3.167
+0.48
3.491
−0.27
−0.565
1.111
3.160
+0.66
3.467
−0.36
−0.552
1.108
3.140
3.480
a
c
c
͑x͒
͑x͒
a0
3.152
3.501
c0
FIG. 3. ͑Color online͒ ͑a͒ Z-contrast image of HfB2 /ZrB2 /Si͑111͒ hetero-
structure. ͑b͒ High resolution XTEM of the perfectly epitaxial HfB2 /ZrB2
interface and corresponding EELS composition profile of Hf ͑M edge͒ and
Zr ͑L edge͒.
For hexagonal films with the ͓0001͔ plane oriented nor-
mal to the substrate, the perpendicular ͑ ͒ and parallel ͑ ͒
c=3.48 Å of bulk HfB2, due to strain imposed by the Si
substrate. These results motivated us to pursue growth of
HfB2 films on isostructural ZrB2 buffers ͓rather than directly
on the Si͑111͒ surface͔ to promote formation of smooth films
suitable for nitride integration. These HfB2 films grow much
more readily ͑ϳ2 nm/min͒ and exhibit exceptional morpho-
logical and structural properties, including flat surfaces
͑AFM roughness ϳ2 nm͒, highly coherent interfaces, and
virtually defect-free microstructures. The XRD measure-
ments show that the layers are partially strained and the lat-
tice parameters are close to those of HfB2 films grown on Si
͑Table I͒. We note that growth of thin HfB2 layers has been
no evidence of epitaxy was given.9 The XTEM data in Fig.
3͑a͒ indicate that our ZrB2 buffers bridge the strain differen-
tial between HfB2 and Si, allowing the formation of perfectly
epitaxial HfB2 films that cannot be obtained directly on Si.
XTEM Z-contrast images and EELS profiles of Zr and Hf
across the interface showed an abrupt transition of the ele-
ments from ZrB2 to HfB2 with no evidence of intermixing
between the two materials at the nanometer scale ͑Fig. 3͒.
In summary, we have shown that heteroepitaxial semi-
metallic HfxZr1−xB2 templates with tunable structural and op-
tical properties represent a potentially very useful practical
system for optoelectronic integration applications on Si.
c
a
strains are given by c=−2C13a/C33, where c=͑c−c0͒/c0
and a=͑a−a0͒/a0. For bulk ZrB2 and HfB2 the known c/a
ratios ͑denoted by below͒ are only slightly different ͑1.114
and 1.108, respectively͒. Therefore, in order to determine the
strain state, we make the following approximations: ͑i͒ for
the relaxed epitaxial film is identical to that of the equilib-
rium bulk crystals and ͑ii͒ the c/a ratio ͑͒ and elastic ratio
=−2C13/C33 are both linear functions of composition;
HfB
ZrB
HfB
ZrB
2
2
2
2
͑x͒=x
+͑1−x͒
and ͑x͒=x
+͑1−x͒ , re-
spectively. Since the elastic constants of both ZrB2 and HfB2
are generally not well known, we used the VASP DFT code7
to calculated them using finite strain deformations of the
equilibrated systems.8 From inversion of the strain relation,
the relaxed lattice constants are then given by a0͑x͒
=͕c͑x͒/͑x͒−͑x͒a͑x͖͒/͕1−͑x͖͒ and c0͑x͒=͑x͒ a0͑x͒, and
these are listed in Table I. We note that the relaxed film
lattice constant of the end members match the known values
for the bulk phases. This provides additional justification for
the approximations discussed above. We also note that the
relaxed lattice constants for the alloys follow Vegard’s law
quite closely. Our analysis shows that ZrB2 and the alloy
films exhibit a slight tensile strain ͑aϳ +0.50%, cϳ
−0.29%͒, while the HfB2 film is strained even more ͑aϳ
+0.66%, cϳ−0.36%͒. The tensile state of these films ͑in-
cluding ZrB2͒ has significant implications for lattice engi-
neering, since this provides better matching with Ga-rich al-
loys. In particular, the measured value of a for the ZrB2 films
is essentially identical to that of GaN.
This work was supported by the NSF ͑EEC-0438400 and
DMR-0221993͒.
1S. S. Ng, Z. Hassan, G. L. Chew, M. R. Hashim, and M. E. Kordesch,
Journal of Physical Sciences 15, 97 ͑2004͒.
As noted above, the in-plane strain systematically in-
creases in the Hf-rich compositional regime with a concomi-
tant reduction in growth rate ͑ϳ0.5 nm/min, for pure HfB2͒.
The increase in strain reduces the reactivity and surface mo-
bility. In fact, for HfB2, the growth on Si͑111͒ eventually
produces almost exclusively rough films with large surface
undulations ͑AFM roughness Ͼ15 nm͒. The growth, in this
case, was conducted at 900 °C via decomposition of pure
Hf͑BH4͒4 under conditions similar to those described above
for the alloys. The RBS and XTEM data confirmed the pres-
ence of predominately rough layers dominated by ensembles
of large islands. Nevertheless, and in spite of the slightly
larger film strain ͑see Table I͒, the data also showed stoichio-
metric and aligned materials with sharp and commensurate
interfaces. XRD off-axis measurements gave a=3.160 Å and
c=3.467 Å which are larger/smaller than a=3.142 Å and
2J. Tolle, R. Roucka, I. S. T. Tsong, C. Ritter, P. A. Crozier, A. V. G.
Chizmeshya, and J. Kouvetakis, Appl. Phys. Lett. 82, 2398 ͑2003͒.
3J. Tolle, J. Kouvetakis, D. W. Kim, S. Mahajan, A. Bell, F. A. Ponce, I. S.
T. Tsong, M. L. Kottke, and Z. D. Chen, Appl. Phys. Lett. 84, 3510
͑2004͒.
4R. Armitage, J. Suda, and T. Kimoto, Phys. Status Solidi C 2, 2191
͑2005͒.
5J. K. Dewhurst, S. Sharma, and C. Ambrosch-Draxl, EXCITING FPLAPW
code, Version 0.9.57, 2006.
6C. Ambrosch-Draxl and J. O. Sofo, Comput. Phys. Commun. 175, 1
͑2006͒.
7G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 ͑1996͒; Comput.
Mater. Sci. 6, 15 ͑1996͒.
8L. Fast, J. M. Wills, B. Johannson, and O. Eriksson, Phys. Rev. B 51,
17431 ͑1995͒.
9S. Jayaraman, J. E. Gerbi, Y. Yang, D. Y. Kim, A. Chatterjee, P. Bellon, G.
S. Girolami, J. P. Chevalier, and J. R. Abelson, Surf. Coat. Technol. 200,
6629 ͑2006͒.
129.120.242.61 On: Wed, 03 Dec 2014 09:57:35