Growth mechanism of single crystal nanowires of fcc metals (Ag, Cu, Ni) and hcp metal (Co) electrodeposited

Article abstract of DOI:10.1149/2.034201jes

Using X-ray diffraction (XRD), transmission electron microscopy (TEM) and scanning electron microscopy (SEM), we have studied the growth of the single crystalline nanowires of Ag, Cu, Ni and Co metals in the small pores (50 nm). We find that the preferential growth surface of the single crystalline nanowires is on the atomically rough surfaces such as fcc(110) and hcp(1010). We have proposed a new model to explain the growth of the nanowires of fcc and hcp metals. In this model we argue that the preferential growth should depend on the number of sites for dehydration of hydrated metal ions on a metal surface. The dehydration occurs only at the apex site of a protruding surface atom since the apex site of the protruding surface atom has an enhanced electrical field. The sites for the dehydration on the fcc(110) and hcp(1010) atomically rough planes are in number much larger than those on the fcc(111) and hcp(0001) atomically smooth surfaces, thus leading to the preferential growth on the fcc(110) and hcp(1010)planes.

Full text of DOI:10.1149/2.034201jes

Journal of The Electrochemical Society, 159 (1) K15-K20 (2012)
K15
0
013-4651/2012/159(1)/K15/6/$28.00 © The Electrochemical Society
Growth Mechanism of Single Crystal Nanowires of fcc Metals
Ag, Cu, Ni) and hcp Metal (Co) Electrodeposited
(
z
Ming Tan and Xinqi Chen
College of Physical Science and Technology, Central China Normal University, Wuhan 430079, China
Using X-ray diffraction (XRD), transmission electron microscopy (TEM) and scanning electron microscopy (SEM), we have studied
the growth of the single crystalline nanowires of Ag, Cu, Ni and Co metals in the small pores (50 nm). We find that the preferential
growth surface of the single crystalline nanowires is on the atomically rough surfaces such as fcc(110) and hcp(10 1 0). We have
proposed a new model to explain the growth of the nanowires of fcc and hcp metals. In this model we argue that the preferential
growth should depend on the number of sites for dehydration of hydrated metal ions on a metal surface. The dehydration occurs
only at the apex site of a protruding surface atom since the apex site of the protruding surface atom has an enhanced electrical field.
The sites for the dehydration on the fcc(110) and hcp(10 1 0) atomically rough planes are in number much larger than those on the
fcc(111) and hcp(0001) atomically smooth surfaces, thus leading to the preferential growth on the fcc(110) and hcp(10 1 0)planes.
©
2011 The Electrochemical Society. [DOI: 10.1149/2.034201jes] All rights reserved.
Manuscript submitted July 12, 2011; revised manuscript received September 13, 2011. Published December 8, 2011.
Investigating the growth mechanisms of metal nanowires in the
cylindrical pores of a template is of fundamental and technological
interests. Such research efforts have been performed. For example,
(
Co) metals is on the fcc (110) and hcp(10 1 0) surfaces. The purpose
of the present paper is to elucidate why the growth surface of the
single crystalline nanowires of fcc and hcp metals is on the (110) and
1
Tian et al. observed two interesting results. First, the nanowires of
(
10 1 0) surfaces, respectively.
Au, Ag, and Cu with an fcc structure are single-crystalline with a pre-
ferred [111] orientation under the low deposition potential, and poly-
crystalline under the high deposition potential. Second, the nanowires
of Ni are polycrystalline, which is relatively insensitive to the depo-
sition potential. They proposed that the formation of the single crys-
talline nanowires should be attributed to the larger critical nucleus
formed at the lower deposition potential. The growth of nanowires
along the [111] direction can be accounted for by the observation
that the {111} surface is the energetically most favorable surface for
fcc metals. However, other groups obtained different experimental
results. Pan et al. prepared single-crystalline metal nanowires of Ni
Experimental
Preparation of the AAO templates.— The porous anodic alu-
mina oxide (AAO) templates were prepared via a two-step anodiza-
5,10,11
tion procedure.
Prior to anodizing, high-purity aluminum foils
(99.999%) were first degreased in acetone and then were annealed in a
−
5
◦
vacuum of 10 Torr at 500 C for 5 h to remove the mechanical stress,
thus obtaining the aluminum foils with a homogeneous structure over
a large area. The aluminum foils were anodized in a 0.3 M H C O
2
2
4
◦
solution at 2 C for 6 h. After removing the alumina layer formed in
the anodization in a mixture of phosphoric acid (6 wt%) and chromic
acid (1.8 wt%), the aluminum foils were anodized again at the same
conditions as the first step for 12 h. The templates that experienced the
in an AAO template at the high deposition potential (−1.0 V and
2
−
4.0 V), which grow preferentially along the [110] direction. They
argued that this preferential growth was owing to the adsorption of H
ions on the cathode, which stabilized the (110) face. However, several
studies show that the pH value of electrolyte, ranging from 2 to 6, had
no effect on the orientation of Ni nanowires. Therefore, their argu-
ment is not supported by the experimental observations. Wang et al.
also observed the preferential growth of Ni nanowires along the [110]
above two-step anodization were etched in a saturated CuCl solution
2
to remove the remaining aluminum on the back side. The alumina
2
–4
barrier layer was then dissolved in a 5 wt% phosphoric acid solution
◦
at 40 C. In order to deposit metal into the pores of AAO templates, a
gold film was sputtered onto the back side of the AAO templates to
serve as the working electrode.
5,6
direction in an AAO template with pore diameter of 25–70 nm.
Moreover, the nanowires of other fcc metals such as Ag and Cu also
7,8
grow preferentially along the [110] direction.
Electrodeposition of the metal nanowires.— The electrolyte for
Metallic cobalt occurs as two crystallographic structures: hcp and
fcc. The structure at room temperature is hcp. The transition temper-
preparation of Ag nanowires was a mixture of AgNO
BO (45 g/L) solutions, with pH of 2 by adding a HNO
tion. The electrolyte for the preparation of nanowires of Cu and Ni
was a mixture of CuSO (90g/L), H BO (45 g/L) solutions, and
NiSO O (100g/L), NiCl O (20g/L), H BO (45 g/L) so-
· 6H
3
(45 g/L) and
H
3
3
3
solu-
◦
ature between hcp and fcc structures is 450 C. Huang et al. found
that the single crystalline Co nanowires deposited at −1.5 V had
4
3
3
9
an hcp structure, which grow along the[10 1 0] direction. Pan et al.
4
· 6H
2
2
2
3
3
also observed the growth of single crystalline hcp Co nanowires
lutions, respectively, and the pH value of the solution was adjusted
to 2.5 with 1 M H SO . The electrolyte for the preparation of Co
along the[10 1 0] direction at the deposition potentials of −0.4 V and
2
4
−
1.0 V. However, these authors do not explain why the growth of
nanowires of hcp Co is along the [10 1 0] direction.
nanowires was a mixture of CoSO (100 g/L) and H BO (45 g/L) in
4
3
3
aqueous solutions, with pH of 2.5 by adding 1 M H
2
SO
4
solution. Di-
From the brief review given above, we can see two points. First,
a lot of studies show that the growth surface of the single crystalline
nanowires of fcc (Ag, Cu and Ni) and hcp(Co) metals is usually
not on the closely packed surface (atomically smooth surfaces) such
as fcc(111) and hcp(0001) but on the atomically rough surface like
rect current (DC) electrodeposition for formation of nanowires of Ag,
Ni and Cu was performed at the deposition potentials of −0.2 V and
◦
at a room temperature (about 25 C), respectively. The Co nanowires
were prepared at deposited potential of −1.6 V at room temperature.
fcc(110) and hcp(10 1 0), respectively. Second, the growth mechanism
of single crystalline metal nanowires is still poorly understood since
the models previously proposed cannot explain why the growth surface
of the single crystalline nanowires of fcc (Ag, Cu and Ni) and hcp
Characterization.— The metal nanowires was characterized
by X-ray diffraction (XRD, Y-2000) with CuK
α
radiation (λ
=
0.154178 nm). The images of deposited metal nanowires were
obtained by scanning electron microscope (SEM, JEOL JSM-6700F)
and transmission electron microscope (TEM, Tecnai G20 Philips). For
XRD measurements the film of Au was mechanically polished away.
For SEM observations, the AAO templates were partly dissolved with
z
E-mail: tanming@phy.ccnu.edu.cn
Downloaded on 2014-11-11 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

K16
Journal of The Electrochemical Society, 159 (1) K15-K20 (2012)
(
(
a)
b)
(220)
(220)
(
c)
(220)
Figure 1. SEM images of the top and side (inset) views of highly ordered
AAO template.
(
d)
( 1010)
5
wt% NaOH solution, and then carefully rinsed with deionized water
for several times. For TEM observations, the AAO was completely
dissolved in a 5% NaOH solution, then rinsed with deionized water
several times, and finally dispersed in absolute ethanol by ultrasonic.
Results
30
40
50
60
70
80
2
θ (degree)
Figure 1 gives a top-view SEM image of AAO template, indicating
that the pores on AAO template are hexagonally arranged and highly
ordered. The average diameter and spacing of pores are estimated to
be 50 and 100 nm, respectively. The inset of Figure 1 is a side-view
image of AAO template, demonstrating that these cylindrical pores
are perpendicular to the AAO template surface.
Figure 2. XRD patterns of metal nanowires deposited at different deposition
potentials: (a) fcc Ag, −0.2 V; (b) fcc Cu, −0.2 V; (c) fcc Ni, −0.2 V (d) hcp
Co, −1.6 V.
Figure 2 shows the XRD patterns of Ag, Cu, Ni and Co nanowires,
which were collected from the top side of nanowires. The nanowires
of Ag, Cu and Ni were prepared at a room temperature under the
negative deposition potential of 0.2 V. The position of the peak of Ag,
Cu and Ni are in good agreement with those of the face-centered cubic
atomically rough surface like fcc(110) and hcp(10 1 0). Here we will
explain why the growth surface of single crystalline nanowires is on
the fcc(110) and hcp(10 1 0).
n+
In the electro-deposition of metal, a metal ion M is transferred
(
0
fcc) Ag (JPCDS, 04-0783), Cu (JPCDS, 04-0836) and Ni (JPCDS,
4-0850), respectively. The sharp peak is due to (220) plane diffrac-
tion. The above XRD results of Ag, Cu and Ni nanowires have pre-
from solution into the ionic metal lattice, meanwhile electrons are pro-
vided from the external electron source (power supply) to the electron
13
gas of the metal M. This electro-deposition process can be pictured
as the following four steps in the atomic-level scale. First, hydrated
metal ions in an aqueous solution diffuse to a metal surface and are
adsorbed on this surface. (In a hydrated metal ion, the water molecules
of hydration are electrostatically attached to the metal ion.) Second,
when an adsorbed hydrated metal ion captures electrons from the sur-
face by quantum-mechanical tunneling, the metal ion is neutralized.
The electrostatic attractive interaction between the neutral metal atom
and water molecules is zero and then the water molecules of hydra-
tion are displaced. Third, the neutral metal atom is adsorbed on the
surface. Fourth, the adsorbed metal atom diffuses to a surface site
viously been observed by other researchers.2
,5,7,12
Using HR-TEM,
those researchers have demonstrated that these nanowires with only
2,12
one diffraction peak is single crystalline. The nanowires of Co were
prepared at −1.6 V. The nanowires of Co prepared at −1.6 V have a
sharp peak at 2θ = 41.6 degrees. Our XRD results of Co nanowires are
9
in agreement with those obtained by Huang et al. Using HR-TEM,
Huang et al. demonstrated that the nanowires of Co with one peak are
9
single crystalline. The peak at 2θ = 41.6 degrees are due to the hcp
9
(
10 1 0) plane diffraction. From the above results, we can see that the
growth surface of single crystalline nanowires is on the atomically
(
such as kink site) where it incorporates into the ionic metal lattice,
rough surface like fcc(110) and hcp(10 1 0).
thus leading to the growth on the surface on which the dehydration
occurs. In the four steps described above, the second step plays a cru-
cial role in the growth of metal nanowires. This is because the metal
nanowires can grow only when the water molecules that are attached
to the metal ions are displaced on a metal surface. The dehydration
involves electron transfer from a metal surface to hydrated metal ions
Figure 3 shows SEM (Fig. 3a) and TEM (Fig. 3b) images of
nanowires of Ag. We can see from Figure 3 that the growing surface
of nanowires is not a completely flat surface but rather a protruding
surface at the nanometer scale.
14
Discussion
by tunneling. (see page 8 of this reference). In the following, we
argue that the probability of electron tunneling is much higher on the
The XRD experimental results given above are not new since these
results have previously been observed. However, we are the persons
who have for the first time taken into consideration all the results
of single crystalline nanowires of fcc and hcp metals and found that
the growth surface of single crystalline nanowires is usually not on
the atomically smooth surface like fcc(111) and hcp(0001) but on the
atomically rough surfaces such as fcc(110) and hcp(10 1 0) than on
the atomically smooth surfaces like fcc(111) and hcp(0001) and so
the dehydration rate is faster on the fcc(110) and hcp(10 1 0) surfaces
than on the fcc(111) and hcp(0001) surfaces. This means that the
preferential growth surface is on the fcc(110) and hcp(10 1 0).
Downloaded on 2014-11-11 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 159 (1) K15-K20 (2012)
K17
Figure 3. SEM(a) and TEM(b) images of Ag
nanowires.
Downloaded on 2014-11-11 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

K18
Journal of The Electrochemical Society, 159 (1) K15-K20 (2012)
Figure 4. Schematic surface structure of fcc and hcp metals, not drawn to scale: (a) fcc(100); (b) fcc(111) and hcp(0001); (c) fcc(110) and hcp(10 1 0), if the
distance between two atoms along the close packed direction is a, then the distance between two atoms is 1.414a along the ꢀ100ꢁ direction and 1.633a along the
ꢀ
0001ꢁ direction, respectively. The fcc(001), fcc(111) and hcp(0001) surfaces are relatively smooth at the atomic scale. However, the fcc(110) and hcp(10 1 0)surface
is atomically rough with the troughs.
Little is known on the tunneling processes occurring in the de-
hydration. In order to understand the tunneling processes in the de-
hydration, we first describe briefly the tunneling processes in field
ion microscopy (FIM), (See page 11 of this reference) which is well
understood. In the presence of a high electrical field of a few V/Å,
an electron in an atom near a metal surface can transfer to the metal
surface by tunneling. The higher the electrical field is, the higher the
probability of tunneling is (this is because the width of tunneling bar-
rier is reduced by the electrical field). In FIM, the metal is applied a
positive bias voltage, the electron of the image gas atom tunnels into
15
Downloaded on 2014-11-11 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 159 (1) K15-K20 (2012)
K19
Figure 5. Schematic illustration of a possible growth mechanism for the single crystalline metal nanowires. Atom 1, originating from dehydration at terrace,
diffuses to a kink site for growth; atom 2 originating from dehydration at step either goes to the layer immediately below and then diffuses along step to a kink site
to join the lattice, or diffuses to a kink site to join the lattice on the same layer; atoms at the topmost layer of growing surface will form a new layer by diffusing.
the metal surface. In the case of electro-deposition, a metal surface
is biased a negative voltage. The value of overpotential should be on
the order of a few V with respect to adsorbed hydrated metal ions,
where it incorporates into the metal lattice. Atom 2, originating from
dehydration at step, either goes to the layer immediately below and
then diffuses along step to a kink site to join the lattice, or diffuses to
a kink site to join the lattice on the same layer. Atoms at the topmost
layer of growing surface of the nanowires can form a new layer by
diffusing. It can be seen from Fig. 5 that the growth can occur domi-
13
which have a zero potential. If the surface metal atom and adsorbed
hydrated metal ion are regarded as the hard balls, the distance between
◦
the centers of them is several A. Therefore there is a high electrical
nantly on the fcc(110) and hcp(10 1 0) surfaces since the fcc(110) and
field between the surface metal atom and adsorbed hydrated metal
14
ion. (see page 5 of this reference) It is the high electrical field that
leads to adsorption of hydrated metal ion on a metal surface and then
electron tunneling from the metal surface to the adsorbed hydrated
metal ion.
hcp(10 1 0) surfaces have an overwhelmingly number of the sites for
the dehydration, compared to the fcc(111) and hcp(0001) surfaces.
SEM and TEM results show that the growing surface of a metal
nanowire is a protruding surface at the nanometer scale, indicating
that the growing surface has a lot of steps. Such protruding surface
has more sites for dehydration than flat surface and therefore favors
the growth of nanowires.
The electrical field is never uniform everywhere as a jellium model
of metal surface would imply. Instead, the field is enhanced near a
protruding atom.15 (See page 73 of this reference) The protruding
atoms may include atoms located at the sites of step, kink and surface
edge. In FIM, a tunneling event occurs only at the apex site of a
The data of workfunctions of single crystal metal surfaces corrob-
orate the above argument that the electron tunneling is much easier on
the atomically rough surface than on the atomically smooth surface.
The value of workfunction represents the height of energy barrier
for electron departing a metal surface. The value of workfunction is
related to probability of electron tunneling. The larger the value of
workfunction is, the lower the probability of electron tunneling is. For
a metal, the value of workfunction is usually lower on the atomically
15
protruding atom and only protruding atoms are imaged. (See Fig.
.3 of this reference) In the case of electro-deposition, therefore, we
1
suggest that the dehydration, which is an electron tunneling process,
should occur only at the apex site of a protruding atom.
Fig. 4 shows the surface structure of (111), (100), (110),
(0001) and(10 1 0) planes of fcc and hcp metals. We can see from
15
rough surface than on the atomically smooth surface. For exam-
ple, the workfunction of the (110) and (111) surfaces is 4.52 eV and
Fig. 4 that the terrace atoms on the fcc(110) and hcp(10 1 0) sur-
faces have two nearest first-layer atoms while the terrace atoms on
the fcc(100), fcc(111) and hcp(0001) have four and six nearest first-
layer atoms, respectively. Thus, the terrace atoms on the fcc(110) and
1
6
16
4
.74 eV for Ag, 5.04 eV and 5.35 eV for Ni, 4.48 eV and 4.94 eV
17
for Cu, respectively. The workfuction of cobalt single-crystal sur-
faces is not available. It can be seen from the workfunction of metal
that the electron tunneling is much easier on the atomically rough sur-
face than on the atomically smooth surface. Therefore, the dehydration
occurs dominantly on the atomically rough surface.
hcp(10 1 0) surfaces can be also regarded as protruding atoms while
the terrace atoms on the fcc(100), fcc(111) and hcp(0001) are not pro-
truding atoms. This observation is confirmed by FIM experimental
15
results. (See Fig. 1.3, Fig. 4.7 and 4.9 of this reference) The electri-
cal field above smooth planes like fcc(100) and fcc(111) is too uniform
and so field adsorption of image gas atoms and electron tunneling will
not occur above the terrace atoms of these surfaces. Adsorption and
tunneling induced by the field occurs only at the apex sites of edge
atoms of the smooth planes since they are protruding atoms. There-
It is well understood for surface scientists that single neutral metal
atom is very active chemically. It sticks to and grows on any surface
which it contacts. This means that one neutral metal atom cannot select
a surface for growth. However, hydrated metal ions in an aqueous
solution can select the rough surface like fcc(110) and hcp(10 1 0)
for dehydration and grow on this surface. This is because the rough
surfaces can make dehydration happen much more easily. Therefore
it is this selection that leads to the dominant growth on the surfaces of
fore only edge atoms are imaged. For example, only edge atoms are
imaged for the nearly circular shape of the (100) layers of an Ir tip.15
(
See Fig. 4.9 of this reference) However, for the nearly circular shape
fcc(110) and hcp(10 1 0).
of the (110) layers of a Pt tip, all the atoms on the (110) surface in-
15
The growth of crystals in nanowires can be influenced by the pore
size of a template. When the pore size is large, several critical nuclei
would form on the Au (or deposited metal) surface of one pore and
polycrystalline growth occurs. In this case, one can expect to observe
the diffraction peaks of not only the (220) plane but also the (111)
and (100) planes. Wang et al. observed the diffraction peaks of the
cluding the terrace atoms are observed in the FIM image. (See Fig.
.7 of this reference) These FIM results indicate that all the atoms on
4
the (110) plane including terrace atoms can be regarded as protruding
atoms. Therefore we believe that all the atoms on the (110) and(10 1 0)
surfaces can be regarded as protruding atoms. On the other hand, for
the (111), (100) and (0001) surfaces, only atoms located at edges and
surface defects such as steps and kink are protruding atoms. Thus, the
sites for the dehydration of hydrated metal ions are in number much
(
111), (220) and (100) planes on the nanowires of Ni, which grew in
5,6
the larger pores of 90–225 nm.
larger on the (110) and (10 1 0) surfaces than on the (111) and (0001)
surfaces.
Fig. 5 schematically gives a possible growth mechanism for the
single crystalline metal nanowires. Atom 1, originating from dehy-
Conclusions
This work has two interesting points. First, we find that the dom-
inant growth of single crystalline nanowires usually occurs on the
dration at terrace of fcc(110) and hcp(10 1 0), diffuses to a kink site
Downloaded on 2014-11-11 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

K20
Journal of The Electrochemical Society, 159 (1) K15-K20 (2012)
4
. H. Pan, B. H. Liu, J. B. Yi, C. Poh, S. H. Lim, J. Ding, Y. P. Feng, C. H. A. Huan,
atomically rough planes such as fcc(110) and hcp(10 1 0) rather than
the atomically smooth planes like fcc(111) and hcp(0001). Second,
we suggest that the growth on a metal surface should be controlled
by the number of sites for dehydration of hydrated metal ions on the
metal surface. The dehydration occurs only at the apex site of a pro-
truding surface atom. The sites for the dehydration on the fcc(110) and
and J. Y. Lin, J. Phys. Chem. B, 109, 3094 (2005).
. X. W. Wang, G. T. Fei, X. J. XU, Z. Jin, and L. D. Zhang, Journal of Physical
Chemistry B, 109, 24326 (2005).
5
6. X. W. Wang, G. T. Fei, L. Chen, X. J. Xu, and L. D. Zhang, Electrochemical and
Solid-State Letters, 10, E1 (2007).
7. X. J. Xu, G. T. Fei, W. H. Yu, L. Chen, and L. D. Zhang, Applied Physics Letters, 88,
211902 (2006).
hcp(10 1 0) rough planes are in number much larger than those on the
fcc(111) and hcp(0001) smooth planes. Therefore the growth of single
crystalline nanowires of fcc and hcp metals can occur dominantly on
8. K. Kim, M. Kim, and S. M. Cho, Materials Chemistry and Physics, 96, 278
(2006).
9
. X. H. Huang, L. Li, X. Luo, X. G. Zhu, and G. G. Li, J. Phys. Chem. C, 112, 1468
2008).
10. B. Wang, G. T. Fei, M. Wang, M. G. Kong, and L. D. Zhang, Nanotechnology, 18,
65601 (2007).
(
the fcc(110) and hcp(10 1 0) planes.
3
1
1
1. H. Masuda and K. Fukuda, Science, 268, 1466 (1995).
2. H. Pan, B. H. Liu, J. B. Yi, C. K. Poh, S. H. Lim, J. Ding, Y. P. Feng, C. H. A. Huan,
and J. Y. Lin, J. Phys. Chem. B, 109, 3094 (2005).
References
13. M. Paunovic and M. Schlesinger, Fundamentals of Electrochemical Deposition
1
2
3
. M. L. Tian, J. G. Wang, J. Kurtz, T. E. Mallouk, and M. H. W. Chan, Nano Letters,
, 919 (2003).
. H. Pan, H. Sun, C. K. Poh, Y. P. Feng, and J. Y. Lin, Nanotechnology, 16, 1559
2005).
. I. Z. Rahman, K. M. Razzeeb, M. A. Rahman, and M. Kamruzzaman, J. Magn. Magn.
Mater., 262, 166 (2003).
(Wiley, New York, 1998).
3
14. S. K. Lower, Electrochemistry (Simon Fraser University, Textbook).
15. T. T. Tsong, Atom-probe field ion microscopy (Cambridge University Press,
Cambridge, New York, 1990).
16. H. B. Michaelson, Journal of Applied Physics, 48, 4729 (1977).
17. P. O. Gartland, S. Berge, and B. J. Slagsvold, Phys. Rev. Lett., 28, 738 (1972).
(
Downloaded on 2014-11-11 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).