Structural phase transitions in Zn(CN)2 under high pressures

Article abstract of DOI:10.1016/j.jssc.2008.09.026

High pressure behavior of zinc cyanide (Zn(CN)₂) has been investigated with the help of synchrotron-based X-ray diffraction measurements. Our studies reveal that under pressure this compound undergoes phase transformations and the structures of

Full text of DOI:10.1016/j.jssc.2008.09.026

ARTICLE IN PRESS
Journal of Solid State Chemistry 182 (2009) 136–140
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Structural phase transitions in Zn(CN)2 under high pressures
Ã
H.K. Poswal a, , A.K. Tyagi b, Andrea Lausi c, S.K. Deb a, Surinder M. Sharma a
a High Pressure Physics Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, India
b Chemistry Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, India
c Synchrotron Trieste-Elettra, S. S. -14 Km 163, 5 in Area Science Park, 34012 Basovizza, Trieste, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
High pressure behavior of zinc cyanide (Zn(CN)2) has been investigated with the help of synchrotron-
based X-ray diffraction measurements. Our studies reveal that under pressure this compound
undergoes phase transformations and the structures of the new phases depend on whether the
pressure is hydrostatic or not. Under hydrostatic conditions, Zn(CN)2 transforms from cubic to
orthorhombic to cubic-II to amorphous phases. In contrast, the non-hydrostatic pressure conditions
drive the ambient cubic phase to a partially disordered crystalline phase, which eventually evolves to a
substantially disordered phase. The final disordered phase in the latter case is distinct from the
amorphous phase observed under the hydrostatic pressures.
Received 13 June 2008
Received in revised form
25 September 2008
Accepted 29 September 2008
Available online 17 October 2008
Keywords:
High pressure
X-ray diffraction
& 2008 Elsevier Inc. All rights reserved.
Structural phase transition
Negative thermal expansion
1. Introduction
contribute to NTE [9], transform to other crystalline phases at 0.2,
0.6 GPa and over 0.7–2.5 GPa, respectively ([4] and reference
therein). On the other hand, cubic ZrV2O7, wherein only the
rotation of the VO4 contributes to NTE, transforms from cubic to
an orthorhombic phase over ꢀ1.4–1.7 GPa [9]. A large number of
anisotropic compounds (Sc2(WO4)3, Lu2(WO4)3, Al2(WO4)3, and
molybdates) [4,7,8] have also been studied under pressure with
the help of X-ray diffraction as well as Raman and IR measure-
ments and these too undergo transitions to crystalline phases
before leading to pressure-induced amorphization (PIA) [10].
However, in spite of the fact that several compounds which show
NTE at room pressure are known to become amorphous at high
pressures, such a correlation may be coincidental. This is because
the transformation to another crystalline phase may almost
eliminate the NTE effect, as observed in Zr(WO4)2 [11]. It is also
interesting to note that molecular dynamics simulations have
shown that model network systems have a tendency to become
amorphous under compression [12].
Recently cyanide-based framework compounds have also been
shown to have NTE [13–16]. Of these Zn(CN)2 is the simplest
coordination framework compound having a cubic crystal struc-
ture and an isotropic and large NTE coefficient (Etwice that for
Zr(WO4)2) over a wide range of temperature. Structural analysis
with the help of atomic pair distribution function implies that
with the increase in the temperature, the transverse displacement
of C–N bridge increases [17]. This compound is very flexible and
soft as implied by the existence of low energy lattice vibrations
[14,18–20].
Though the property of negative thermal expansion (NTE) is
known for a long time, this topic got an impetus after the
discovery of a large isotropic NTE in Zr(WO4)2 over a wide range of
temperature [1–5]. Aside from the interesting underlying physics,
the research in this topic is also driven by the technological
applications. For example some applications may require materi-
als with pre-designed and small thermal expansion [6]. Generally
such materials are engineered through an appropriate mixture of
compounds having positive and NTE. In such a composite making
process, grain interactions are subjected to local stresses. Thus it is
desirable that the component compounds of the composite should
not undergo any local stress-induced phase transformation which
may substantially affect the thermal expansion behavior. In this
context, the knowledge of the behavior of NTE compounds under
pressure may be useful.
Many of the compounds displaying NTE are oxides having
framework structures with a strong single-atom bridge, such as in
A(MO4)2, AM2O7, A2(MO4)3 (A ¼ Zr, Hf, Sc, Lu, Al etc., M ¼ W, Mo, V,
etc.) and NASICON families of compounds [4,7,8]. High-pressure
structural studies of these materials have shown that these
undergo phase transitions to other crystalline phases at relatively
low pressures, eventually becoming amorphous at higher pres-
sures. For example, cubic Zr(WO4)2, Hf(WO4)2 and g-Zr(MoO4)2,
where the librational motion of both the constituent polyhedra
The structure of Zn(CN)2, shown in Fig. 1, consists of two
interpenetrating three dimensional tetrahedral framework made
up of Zn–C–N–Zn chains. Zinc cyanide (Zn(CN)2) may have either
Ã
Corresponding author. Fax: +91222 550 5151.
E-mail addresses: himanshu@barc.gov.in, poswalhimanshu@gmail.com
(H.K. Poswal).
0022-4596/$ - see front matter & 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2008.09.026

ARTICLE IN PRESS
H.K. Poswal et al. / Journal of Solid State Chemistry 182 (2009) 136–140
137
The experiments were performed in hydrostatic as well as non-
hydrostatic pressure environments. For the studies under hydro-
static pressures, methanol: ethanol: water mixture (16:3:1) was
used as a pressure transmitting fluid which provides almost
hydrostatic conditions up to ꢀ15 GPa [16]. For the experiments
under non-hydrostatic conditions, no pressure transmitting fluid
was used. The pressure was determined from the known equation
of state of gold [24]. High-pressure angle dispersive X-ray-
diffraction experiments were carried out at the 5.2 R (XRD1)
beamline of Elettra synchrotron source, with monochromatic
˚
X-rays of wavelength 0.67019 A. The diffraction patterns were
recorded using MAR345 imaging plate detector kept at a distance
of ꢀ200 mm from the sample. Two-dimensional imaging plate
records were transformed to one dimensional diffraction profiles
by the radial integration of diffraction rings using the FIT2D
software [25].
3. Results and discussion
Fig. 2 shows one dimensional angle dispersive X-ray diffraction
profiles of Zn(CN)2 at a few representative pressures under
hydrostatic conditions. Analysis of diffraction patterns indicates
that Zn(CN)2 remains in the cubic structure up to ꢀ1 GPa. The
Rietveld refinement carried out on the diffraction pattern
recorded at 0.4 GPa confirms the earlier difficulty in resolving
CN-ordered form ðP43mÞ from the CN-disordered one ðPn3mÞ with
Fig. 1. Crystal structure of Zn(CN)2 at ambient conditions ( color online). Large
(light blue) spheres at the vertices represent Zn. Other two smaller atoms (deep
blue and green) are C and N. In the disordered structure the occupancy of each
equivalent site by C and N is 1/2.
of the two probable cubic structures, the CN-ordered structure
(space group P43m) [15] and CN-disordered structure (space
group Pn3m) [13]. In the ordered structure the C and N ions, lying
along the body diagonal, are ordered such that they form alternate
ZnC4 and ZnN4 tetrahedra. And in the disordered structure,
statistically the site occupancy by C and N atoms is 0.5. Because
of close values of X-ray atomic scattering factors, it is very difficult
to differentiate between these two probable structures using
X-ray diffraction technique. However, the thermal displacement
analysis of neutron time-of-flight diffraction measurements favors
the disordered structure [13]. Subsequent Raman and IR spectro-
scopic studies also support a disordered cubic structure for
Zn(CN)2 [21].
The present study aims to investigate the high pressure
behavior of Zn(CN)2. Aside from the fact that pressure, and the
consequent phase transitions, would affect the NTE behavior, it
would also help identify the features common to other NTE
compounds. Earlier high pressure studies (Pp2 GPa) have pro-
vided some useful information. For example, Chapman and
Chupas [22] have shown that (upto ꢀ1 GPa) pressure enhances
NTE i.e.,
d
a
/
d
P ꢀꢁ5 ꢂ10ꢁ6 Kꢁ1/GPa. Another investigation claims
that this compound becomes almost amorphous at very low
pressures, ꢀ1.6 GPa [23]. In this manuscript, we present the
results of our high pressure angle dispersive X-ray diffraction
measurements on Zn(CN)2 employing synchrotron radiation.
2. Experimental details
Zn(CN)2 sample was synthesized employing precipitation
technique using ZnCl2 and KCN solutions and the product was
characterized to be Zn(CN)2 with the help of X-ray diffraction
measurements. For the high pressure experiments, Zn(CN)2 was
loaded in a tungsten gasket hole of ꢀ120
m
m diameter, drilled
Fig. 2. X-ray diffraction patterns of Zn(CN)2 at a few representative pressures
under hydrostatic conditions. Open circles and asterisks represent the diffraction
peaks from tungsten gasket and gold pressure marker, respectively. h k l indices are
given only for the cubic Zn(CN)2.
after pre-indenting it to a thickness of ꢀ70
mm and mounted in a
Mao-Bell kind of diamond-anvil cell (DAC). For pressure calibra-
tion a few particles of gold were also loaded in the gasket hole.

ARTICLE IN PRESS
138
H.K. Poswal et al. / Journal of Solid State Chemistry 182 (2009) 136–140
the help of X-ray diffraction data. For the initial cubic phase, the
variation of the lattice parameter was obtained by analyzing the
diffraction patterns with the help of Le Bail profile fitting, as
implemented in GSAS software [26]. From a very few P–V data
points obtained in the initial cubic phase, it is difficult to
determine bulk modulus (B) and its derivative (B0). However, we
find that our data are consistent with the results of Ref. [17] (i.e.,
with B ¼ 34.19(21) GPa and B0 ¼ ꢁ6.0(7)). On increase of pressure
beyond 1 GPa many new diffraction peaks at higher d-values
appeared. These observed changes indicate that the structure
has transformed to a new high pressure phase, having larger
lattice parameters. Coexistence of ambient phase with the high
pressure phase implies the transformation to be of the first
order. In order to index the high pressure phase, Crysfire software
[27] was used which gave the new structure to be of orthorhombic
symmetry and with lattice parameters, a ¼ 4.109, b ¼ 7.113 and
˚
c ¼ 13.165 A at 1.3 GPa. Checkcell software [28] was used to
obtain the best possible space group (consistent with extinction
conditions) to be Pmc21. Fig. 3 shows that orthorhombic cell
gives excellent Le Bail profile fitting of the experimental
diffraction pattern. (The details of indexing with this orthorhom-
bic cell are given as the supplementary information.) However,
due to the observation of only a few diffraction peaks, the detailed
structure of the new phase could not be determined. Density
considerations suggest that each unit cell in the orthorhombic
phase may have at least four formula units and this would
imply ꢀ5.9% volume reduction per formula unit across the
phase transition. This is consistent with the multiplicity of sites
in Pmc21 space group.
Fig. 4. Le Bail profile fitting of the observed diffraction pattern of Zn(CN)2 at
2.5 GPa employing three phases i.e., gold pressure marker, tungsten gasket and
cubic high pressure phase.
The diffractions patterns in Fig. 2 imply that the structure
evolved to still another phase at ꢀ2 GPa. This could be indexed
(using Crysfire software [27]) to be a cubic phase with a rather
˚
large lattice parameter, i.e., a ¼ 12.669 A at 2.5 GPa. As displayed
in Fig. 4, three phase Le Bail profile refinement (i.e., with gold
pressure marker, tungsten gasket and high pressure cubic phase)
shows a very good fitting to the experimental data. (Details of this
indexing too are submitted as supplementary information.) Here
too, due to the limited number of observed diffraction peaks, the
structure could not be determined. However, the density
considerations suggest that this large unit cell should have at
least 22 formula units per unit cell and this would imply at least
Fig. 5. Variation of volume per formula-unit with pressures under hydrostatic
conditions in three different phases, solid squares represent ambient cubic phase
(with formula units/cell (Z ¼ 2)), solid diamond represents high pressure
orthorhombic phase (Z ¼ 4) and solid triangles represent high pressure cubic
phase (Z ¼ 22).
ꢀ4% volume reduction per formula unit across the phase
transition. Extinction conditions give two probable space groups
viz., P213 or P4232. However, multiplicities of site positions of P213
space group indicate that there should be 20 formula units per
unit cell. This will lead to higher volume per formula units
compared to the low pressure phase, which is unphysical. These
observations suggest that high pressure cubic phase may belong
to P4232 space group. This space group can also rationalize the
existence of 22 formula units per unit cell. This high pressure
phase remained stable up to ꢀ9 GPa. The variation of volume/
formula-unit, as deduced from the diffraction data up to ꢀ9 GPa, is
shown in Fig. 5.
Beyond ꢀ9 GPa, all the diffraction peaks started showing
broadening concomitant with the decrease in the intensities.
The observation of a broad hump at 2
yꢀ141 implies that the
Fig. 3. Le Bail profile fitting of the diffraction pattern of Zn(CN)2 at 1.3 GPa with
four phases i.e., gold pressure marker, tungsten gasket, orthorhombic high
pressure phase and ambient cubic phase. The coexistence of the parent and the
daughter phases implies the phase transition at ꢀ1 GPa to be of first order.
structure becomes increasingly disordered at higher pressures. By
ꢀ11 GPa almost all the diffraction peaks of the sample vanished
leaving behind only the hump and the diffraction peaks from the

ARTICLE IN PRESS
H.K. Poswal et al. / Journal of Solid State Chemistry 182 (2009) 136–140
139
˚
pressure marker (gold) and the gasket (tungsten). This indicates
that Zn(CN)2 transforms to an amorphous phase at ꢀ11 GPa. To
test whether this amorphous phase transforms to any crystalline
phase if overdriven, pressure was further raised up to ꢀ14 GPa.
However, our results show that it remained amorphous. On
release of pressure the diffraction pattern did not show any
reappearance of the crystalline structure, establishing irreversible
PIA of Zn(CN)2.
2.68 A. Due to the broad nature of these peaks, the diffraction
pattern could not be indexed [30]. However, careful analysis
shows that the broad peaks under non-hydrostatic pressures are
not the overlapped broadened peaks of hydrostatic experiments.
Thus the new phase observed under non-hydrostatic stresses is
distinct from that obtained under hydrostatic conditions. At still
higher pressures, the diffraction peaks became very broad and our
data suggests that by ꢀ10.5 GPa the sample became substantially
disordered and remained so upto ꢀ11.7 GPa. On release of
pressure, the sample remained in the disordered state i.e., did
not revert to any crystalline phase, much like in the hydrostatic
experiments. We should note that the disordered phase under
non-hydrostatic conditions still has some persistent diffractional
features at ꢀ11.7 GPa, at variance with the results obtained under
hydrostatic conditions where the sample became essentially
amorphous at ꢀ11 GPa. These results imply that under non-
hydrostatic conditions, the structure may not be as disordered as
under hydrostatic pressures and this difference may have its roots
in the fact that the preceding structures are different in these two
cases [31]
Several high pressure studies have earlier been carried
out on various cyanide compounds [32–34]. Raman investi-
gations on K2Hg(CN)4 show several phase transitions, proposed
to be related to the modifications of tetrahedra as well as
bending of C–Hg–C and Hg–C–N angles [32]. Hg(CN)2 has
substantially different structure but still undergoes several
high pressure phase transitions [33]. As mentioned in Section 1,
earlier high pressure investigations on Zn(CN)2 suggested
PIA of this compound at ꢀ1.6 GPa [23]. However, due to the
usage of laboratory X-ray source, the diffraction pattern had
poor signal/noise ratio and hence the present study supersedes
those results.
Before we present the results under non-hydrostatic pressure
conditions, it would be useful to state here a few general points.
As the non-hydrostatic conditions in a DAC are necessarily due to
the heterogeneous stress distribution, it is likely that the
diffraction peaks would be broader in this case. While, in
principle, non-hydrostatic stresses can change the nature of phase
transformation ([4,5,29] and reference therein), one should also
take care of the fact that the heterogeneity in stress distribution
can bring about multiple phase co-existence. However, when the
size of the grains in the powdered sample is orders of magnitude
smaller than the dimensions of the sample contributing to
diffraction pattern, one may expect a reasonable degree of
reproducibility in the experimental results. With these general
points, we present the diffraction patterns for Zn(CN)2 under non-
hydrostatic pressure conditions in Fig. 6. In this case the initial
cubic phase persisted up to ꢀ2 GPa, indicating somewhat
extended stability region of the cubic phase than under the
hydrostatic pressures. On further increase of pressure diffraction
peaks continued to broaden. At ꢀ3.5 GPa diffraction pattern
showed new, but broad, diffraction peaks at 4.75, 4.28, 3.65 and
The studies carried out so far do not elucidate the mechanism
of PIA of Zn(CN)2. We may mention that in some materials such as
in ice and quartz, softening of a transverse acoustic branch (and
hence an appropriate elastic constant) has been proposed to be
the mechanism for the PIA [35–37]. In such cases amorphous
phase may be viewed as an assemblage of a large number of
incommensurate states [38]. Such a mechanism for the amorphi-
zation would imply that this transformation should get completed
over a very small range of pressure, if not at a specific value of the
pressure. For Zn(CN)2, the amorphous/disordered phases arise
over a pressure range of ꢀ2 GPa, suggesting that softening of
elastic constants may not be the mechanism for Zn(CN)2. In any
case, for such a mechanism for Zn(CN)2, the computed elastic
modulus must be shown to vanish in the phase just preceding the
amorphous phase, necessitating its unambiguous structural
determination. A more general interpretation of PIA is that the
amorphous state is a kinetically preferred state brought about by
the inability of the compound to reach a higher coordinated
crystalline state [10]. A recent investigation on Zr(WO4)2 suggests
a similar mechanism is operative even for other compounds
showing NTE (which have a large number of low frequency
modes) [39]. This should encourage further studies on Zn(CN)2 at
high pressures and temperatures.
In conclusion, under the hydrostatic pressures, we have
observed three phase transformations i.e., cubic-orthorhombic
-cubic-II-amorphous. PIA is observed to occur at
ꢀ11 GPa, at variance with the results published earlier
where it was claimed to take place at ꢀ1.6 GPa [21,23]. In contrast,
under non-hydrostatic conditions, Zn(CN)2 undergoes two
structural transformations i.e., cubic phase to a partially dis-
ordered crystalline phase followed by an evolution to a substan-
tially disordered phase. Moreover, the disordered phase in
this case is different from that obtained under hydrostatic
conditions.
Fig. 6. X-ray diffraction patterns of Zn(CN)2 at a few representative pressures
under non-hydrostatic conditions. Open circles and asterisks represent the
diffraction peaks from tungsten gasket and gold pressure marker, respectively.

ARTICLE IN PRESS
140
H.K. Poswal et al. / Journal of Solid State Chemistry 182 (2009) 136–140
Acknowledgment
[17] K.W. Chapman, P.J. Chupas, C.J. Kepert, J. Am. Chem. Soc. 127 (2005)
15630.
[18] K.W. Chapman, M. Hagen, C.J. Kepert, P. Manuel, Physica B 385 (2006) 60.
[19] A.L. Goodwin, C.J. Kepert, Phys. Rev. B 71 (2005) 140301.
[20] A.L. Goodwin, Phys. Rev. B 74 (2006) 134302.
Authors are thankful to Dr. Nandini Garg for her help during
experiments.
[21] T.R. Ravindran, A.K. Arora, T.N. Sairam, J. Raman Spectroscopy 38 (2007) 283.
[22] K.W. Chapman, P.J. Chupas, J. Am. Chem. Soc. 129 (2007) 10090.
[23] T.R. Ravindran, A.K. Arora, S. Chandra, M.C. Valsakumar, N.V. Chandra Shekar,
Phys. Rev. B 76 (2007) 054302.
Appendix A. Supporting Information
[24] A. Dewaele, P. Loubeyre, M. Mezouar, Phys. Rev. B 70 (2004) 94112.
[25] A.P. Hammersley, S.O. Svensson, M. Hanfland, A.N. Fitch, D. Hauserman, High
Pressure Res. 14 (1996) 235.
[26] A.C. Larson, R.B. Von Dreele, Los Alamos National Laboratory Report LAUR, 86,
2000.
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.jssc.2008.09.026.
[27] R. Shirley, Crysfire, 2004.
[28] Jean Laugier et Bernard Bochu /http://www.inpg.fr/LMGPS or /http://www.
ccp14.ac.uk/tutorial/lmgp/S.
References
[29] N. Garg, S.M. Sharma, J. Phys.: Condens. Matter 19 (2007) 456201.
[30] The observed pressure induced variations of the d-spacings under non-
hydrostatic pressures and their comparison with the corresponding varia-
tions under hydrostatic pressures has been submitted as supplementary
information.
[1] J.S.O. Evans, T.A. Mary, T. vogt, M.A. Subramanian, A.W. Sleight, Chem. Mater. 8
(1996) 2809.
[2] M.G. Tucker, A.L. Goodwin, M.T. Dove, D.A. Keen, S.A. Wells, J.S.O. Evans, Phys.
Rev. Lett. 95 (2005) 255501.
[3] A.W. Sleight, Curr. Opin. Solid State Mater. Sci. 3 (1998) 128.
[4] S.K. Sikka, J. Phys.: Condens. Matter 16 (2004) S1033.
[5] G.D. Barrera, J.A.O. Bruno, T.H.K. Barron, N.L. Allan, J. Phys.: Condens. Matter
17 (2005) R217.
[6] C.N. Chu, N. Saka, N.P. Suh, Mater. Sci. Eng. 95 (1987) 303.
[7] N. Garg, C. Murli, A.K. Tyagi, S.M. Sharma, Phys. Rev. B 72 (2005) 064106.
[8] N. Garg, V. Panchal, A.K. Tyagi, S.M. Sharma, J. Solid Stat. Chem. 178
(2005) 998.
[31] In principle, one cannot rule out the possibility of pressure medium (alcohol
mixture) reacting irreversibly with Zn(CN)2 at high pressures. However, we
feel that it is unlikely as at low pressures the X-ray diffraction patterns as well
as Raman spectra are same as that of the initial cubic phase, and visual
observations show that at high pressures the nature of powdered sample
stays the same as loaded initially. In addition, at higher pressures the new
phases are expected to be denser than the initial phases.
[32] P.T.T. Wong, Phys. Rev. B 23 (1981) 375.
[9] S. Carlson, A.M.K. Anderson, J. Appl. Cryst. 34 (2001) 7.
[10] S.M. Sharma, S.K. Sikka, Prog. Mater. Sci. 40 (1996) 1.
[11] J.S.O. Evans, Z. Hu, J.D. Jorgensen, D.N. Argyriou, S. Short, A.W. Sleight, Science
275 (1997) 61.
[12] R.J. Speedy, J. Phys.: Condens. Matter 8 (1996) 10907.
[13] D. Williams, D.E. Parting, F.J. Lincoln, J. Kouvetakis, M. O’Keeffe, J. Solid State
Chem. 134 (1997) 164.
[14] J.W. Zwanziger, Phys. Rev. B 76 (2007) 52102.
[15] B.F. Hoskins, R. Robson, J. Am. Chem. Soc. 112 (1990) 1546.
[16] R.J. Angel, B. Maciej, Z. Jing, G. Diego, S.D. Jacobson, J. Appl. Cryst. 40
(2007) 26.
[33] D.M. Adams, P.D. Hatton, J. Phys. C: Solid Stat. Phys. 16 (1983) 3349.
[34] G. Dehnicke, K. Dehnicke, H. Ahsbahs, E. Hellner, Ber. Bunsen. Phys. Chem. 78
(1974) 1010.
[35] Th. Stra¨ssle, A.M. Saitta, S. Klotz, M. Braden, Phys. Rev. Lett. 93 (2004) 225901.
[36] J.S. Tse, D.D. Klug, C.A. Tulk, I. Swainson, E.C. Svensson, C.K. Loong, V. Shpakov,
V.R. Belosludov, R.V. Belosludov, Y. Kawazoe, Nature (London) 400 (1999) 647.
[37] K. Umemoto, R.M. Wentzcovitch, S. Baroni, S. Gironcoli, Phys. Rev. Lett. 92
(2004) 105502.
[38] S.M. Sharma, Indian J. Pure Appl. Phys. 33 (1995) 562.
[39] D.A. Keen, A.L. Goodwin, M.G. Tucker, M.T. Dove, J.S.O. Evans, W.A. Crichton,
M. Brunelli, Phys. Rev. Lett. 98 (2007) 225501.