Preparation and microwave properties of Ni hollow fiber by electroless plating-template method

Article abstract of DOI:10.1016/j.jallcom.2007.05.016

Nickel-coated silk composites (Ni/silk fiber) were prepared by electroless nickel-plating using a kind of natural silk as template in first stage. The silk templates were then removed to get Ni hollow fiber by annealing the Ni/silk fiber at 220 °C for 2 h in air or 700 °C for 1.5 h in nitrogen. The prepared Ni/silk fiber and Ni hollow fiber were characterized by X-ray diffraction (XRD), scanning electron microscope (SEM), field emission scanning electron microscope (FE-SEM) and differential thermal analysis and thermogravimetric analysis (TG-DTA). The values of complex permittivity, complex permeability, dielectric and magnetic loss of Ni/silk fiber and Ni hollow fiber were measured in the frequency range of 2-18 GHz by a reflection/transmission approach using a network analyzer. The results indicate that the dielectric loss of Ni/silk fiber and Ni hollow fiber is high up to 1 even at 18 GHz while the magnetic loss is low.

Full text of DOI:10.1016/j.jallcom.2007.05.016

Journal of Alloys and Compounds 458 (2008) 588–594
Preparation and microwave properties of Ni hollow fiber
by electroless plating-template method
∗
Haijun Zhang , Yun Liu
High Temperature Ceramics Institute, Zhengzhou University, 75 Daxue Road, Zhengzhou, Henan Province 450052, PR China
Received 8 February 2007; received in revised form 29 April 2007; accepted 5 May 2007
Available online 10 May 2007
Abstract
Nickel-coated silk composites (Ni/silk fiber) were prepared by electroless nickel-plating using a kind of natural silk as template in first stage.
◦
◦
The silk templates were then removed to get Ni hollow fiber by annealing the Ni/silk fiber at 220 C for 2 h in air or 700 C for 1.5 h in nitrogen.
The prepared Ni/silk fiber and Ni hollow fiber were characterized by X-ray diffraction (XRD), scanning electron microscope (SEM), field emission
scanning electron microscope (FE-SEM) and differential thermal analysis and thermogravimetric analysis (TG-DTA). The values of complex
permittivity, complex permeability, dielectric and magnetic loss of Ni/silk fiber and Ni hollow fiber were measured in the frequency range of
2
–18 GHz by a reflection/transmission approach using a network analyzer. The results indicate that the dielectric loss of Ni/silk fiber and Ni hollow
fiber is high up to 1 even at 18 GHz while the magnetic loss is low.
2007 Elsevier B.V. All rights reserved.
©
Keywords: Metals and alloys; Chemical synthesis; Dielectric response; Magnetisation; Scanning electron microscopy
1
. Introduction
and capability of depositing on either conductive or nonconduc-
tive substrates. Recently, electroless plating has also attracted a
lot of interests in preparation of low dimensional nanostructured
material [14–19] and in the decoration on carbon nanotubes, SiC
ultrafine powder, etc. [17,20–24].
Althoughelectrolessplatingrouteshavebeenpreviouslyused
to produce various composite powders with multiple compo-
nents and coatings, metal hollow fiber prepared by these routes
has not been previously reported. And to the knowledge of the
authors, the microwave properties of hollow nickel fiber have not
been previously reported yet. This paper reports our results on
preparation and microstructure of Ni hollow fiber by an elec-
troless plating and template removing method. The complex
permittivity and permeability of Ni hollow fiber in 2–18 GHz
were also investigated.
Metal fiber is a kind of very important potential candidate for
radar adsorbing materials (RAM) [1–2]. A great deal of work has
been done on both flaky and spherical magnetic metal particles.
The results showed that those particles exhibited magnetic losses
in the microwave frequency range, and their microwave perme-
ability depended on particle size [3–8]. However, there have
been few studies of the microwave characteristics of magnetic
metal fiber. Generally speaking, the methods of bundle-drawing,
melt-extraction and chatter-machining [9–13] can be used for
production of metal fiber more than 10 ␮m. However, it is very
difficult to get hollow metal fiber with the above-mentioned
methods. As a matter of fact, using as RAM, metal hollow fiber
seems to be better than metal solid fiber in terms of its lower
weight.
Electroless plating is widely used in many fields for its advan-
tages, such as low cost, easy formation of a continuous and
uniform coating on the surface of substrate with complex shape
2
. Experimental procedure
A kind of natural silk was used as template to prepare nickel-coated silk
composites (Ni/silk fiber). The original average length of the starting silk is
more than 300 ␮m, as shown in Fig. 1 indicating that the surface of the starting
silk is very smooth. The original silk was cut into pieces with a length of about
50–100 ␮m for electroless plating.
∗
Corresponding author. Tel.: +86 371 67761908; fax: +86 371 67763822.
E-mail addresses: zhanghaijun@zzu.edu.cn, zhjzdl@yahoo.com.cn
H. Zhang).
Surface treatment of the natural silk was carried out before electroless plat-
ing in order to get complete, continuous nickel film. First, the natural silk was
(
0
925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2007.05.016

H. Zhang, Y. Liu / Journal of Alloys and Compounds 458 (2008) 588–594
589
Fig. 1. SEM photograph of natural surface of silk.
immersed in acetone and then in HF (aq) with ultrasonic vibration for 15 min.
is 3.0; the concentration of [Ni2+] is 0.25 M; the molar ratio of [Na3(Cit)]/[Ni2+
]
is 1:4; the bathing time is 2 h; the amount of CTMAB is 0.5 wt.% relative to
final ultrafine nickel powders.
After that, the cleaned natural silk was sensitized in a mixed solution of stan-
nous chloride (SnCl2·H2O) and hydrochloric acid (HCl) solution for 1 h and
then activated in palladium chloride (PdCl2) and hydrochloric acid for another
Differential thermal analysis coupled with thermogravimetric analysis of
the Ni-coated silk composites was carried out on a NETZSCH STA-449C
1
h.
◦
Solution of nickel ion was prepared by dissolving an analytical grade
Thermal Analysis System with a heating temperature rate of 10 C/min in
NiSO4·6H2O in distilled water. The pH values of the solution was adjusted
flowing air. The sample pot was platinum, and the reference material was
␣-alumina.
◦
by sodium carbonate solution to 10.7 and the solution was heated to 80 C.
Cetyltrimethyl ammonium bromide (CTMAB) was chosen as surfactant, with
sodium citrate (Na3(Cit)) as the complex agent, and hydrazine hydrate as the
reducing agent. Electroless nickel-plating solution was prepared by adding the
following components in the following order:
The XRD data were collected by PHILIPS diffractometer (Model X’Pert
◦
◦
PRO, Cu K␣ radiation, 40 kV, 40 mA) in the 2θ ranging from 30 to 80 with
◦
a step width of 0.016 . The apparent crystallite size of the coated nickel was
determined using the Scherrer formula.
Fiber morphology was characterized via scanning electron microscopy
(
SEM) (Model, JSM-5610LV, JEOL, JAPAN, 20 kV) coupled with EDS (Model,
Oxford 6587, England) and field emission scanning electron microscopy
FE-SEM) (Model, JSM-6700F, JEOL, JAPAN, 15 kV). The fiber was ultra-
NiSO4·6H2O
Na3(Cit)
CTMAB
0.25 mol/L
0.0625 mol/L
0.5 wt.% relative to starting natural silk
0.75 mol/L
(
sonically dispersed into water, and the resultant suspension was spread
on the surface of a polished silicon plate. The samples were coated
with a thin layer of platinum for conductivity before SEM and FE-SEM
observation.
Ni/silk fiber–paraffin wax and the Ni hollow fiber–paraffin wax compos-
ites were prepared by homogeneously mixing the fiber with paraffin wax (with
about 20% of fiber by volume). Toroidal-shaped samples with an inner diameter
of 3.02 mm, an outer diameter of 7.0 mm and a length of 3–4 mm were pre-
pared. A network analyzer (Agilent PNAL N5230A Network Analyzers) was
employed to measure reflected and transmitted scattering parameters (S11, S21)
Hydrazine hydrate
The silk was placed in nickel solution for 2 h. Then the mixture composites
were separated from the solution by filtration separation, washed with distilled
water, ethanol and acetone respectively, and dried at 60 C for 12 h. Gray-black
Ni-coated silk composites were obtained. The Ni-coated silk composite was
annealed in air (or nitrogen) atmosphere to remove the silk template, and to get
the Ni hollow fiber.
To compare the micrograph of pure nickel powder with that of the Ni/silk
fiber, pure nickel powders were also prepared from NiSO4 aqueous solution by
electroless plating method using hydrazine hydrate as reductant. The principal
factors of pure ultrafine Ni powder synthesis are as follows: pH value of bathing
◦
ꢀ
ꢀꢀ
ꢀ
ꢀꢀ
and determine the values of ε , ε , μ and μ of Ni/silk fiber and Ni hollow fiber
in the frequency range of 2–18 GHz by using a reflection/transmission technique
[25,26].
◦
2+
solution is 10.7; bathing temperature is 80 C; molar ratio of [N2H4·H2O]/[Ni
]
Fig. 2. Differential thermal analysis and thermogravimetric of Ni/silk in air.
Fig. 3. XRD results of Ni/silk, Ni/silk annealed in air and nitrogen.

5
90
H. Zhang, Y. Liu / Journal of Alloys and Compounds 458 (2008) 588–594
Fig. 4. SEM (FE-SEM) photograph of Ni/silk and nickel powders by electroless plating.
Table 2
The EDS results of Ni hollow fiber in Fig. 5a
Table 1
Spectrum
O
Si
Ni
Total
The EDS results of Ni/silk fiber in Fig. 4c
Spectrum 1
Spectrum 2
Spectrum 3
Spectrum 4
Spectrum 5
2.00
1.85
1.72
0.00
2.17
1.36
1.93
1.46
0.87
1.01
96.64
96.21
96.81
99.13
96.82
100.00
100.00
100.00
100.00
100.00
Spectrum
C
O
Si
3.99
38.89
31.64
Ni
Total
Spectrum 1
Spectrum 2
Spectrum 3
5.95
9.53
4.52
2.24
3.13
1.22
87.82
48.45
62.62
100.00
100.00
100.00
Mean
6.67
2.20
24.84
66.30
Mean
1.55
1.33
97.12
100.00
All results in weight percent.
All results in weight percent.

H. Zhang, Y. Liu / Journal of Alloys and Compounds 458 (2008) 588–594
591
◦
Fig. 5. SEM (FE-SEM) photograph of Ni hollow fiber annealed at 210 C for 2 h in air.
◦
Fig. 6. SEM (FE-SEM) photograph of Ni hollow fiber annealed at 700 C for 1.5 h in nitrogen.

5
92
H. Zhang, Y. Liu / Journal of Alloys and Compounds 458 (2008) 588–594
3
. Results and discussions
fiber. Ni element resulted from the electroless plating and C ele-
ment from the starting silk. The observed silicon element can
be ascribed to the silicon plate bearing the composite powder,
and the oxygen element may be due to little oxidation of coated
nickel. These results confirm that the prepared fiber was Ni/silk
fiber. Comparing Ni/silk fiber (Fig. 4e and f) with the starting
silk (Fig. 1) shows that the surface of Ni/silk fiber is uneven and
composed of small particles with size less than 100 nm (Fig. 4f)
and Ni nanocrystals are arrayed nearly parallel to the axes of the
starting silk (Fig. 4e and f). It can be seen by comparing Fig. 4e
and f with Fig. 4a that the morphologies of coated Ni on the sur-
face of Ni/silk fiber are also different from the pure Ni particles
by electroless plating as the size of coated Ni on the surface of
Ni/silk fiber is much smaller than that of the pure nickel. The
result suggests that the presence of the silk template impedes the
grain-boundary movement of the coated nickel and plays a role
in preventing grain-growth.
The TG-DTA curves of the prepared Ni/silk fiber in air were
shown in Fig. 2. The results indicated that the weight change
could be divided into two temperature regions, namely, from
◦
room temperature (RT) to about 340 C with a weight loss of
◦
about 5% and from 340 to 500 C with a weight increase of about
2% to initial weight. At temperature higher than 500 C, there
◦
1
was almost no change in weight increase that can be attributed to
the formation of a pure oxide system (NiO). In the temperature
◦
region from RT to about 340 C, the weight loss can be ascribed
to the oxidation of the template silk playing the main role. The
◦
weight gain in 340–500 C might result from the oxidation of
the coated nickel. Based on the above results, the Ni-coated silk
◦
◦
composite was annealed at 220 C for 2 h in air (or at 700 C for
.5 h in nitrogen atmosphere) in order to avoid the oxidation of
1
the coated nickel and obtain Ni hollow fiber.
It is seen from Fig. 3 that only pure Ni was present in all
samples of original Ni/silk fiber, and the Ni/silk fiber annealed
in air and nitrogen, and the crystalline size of coated nickel
was about 20–30 nm as determined by the XRD line-broadening
technique. These results indicate that nickel was coated on the
surface of the natural silk, and no oxidation of the coated nickel
occurred during the heat treatment process either in air or in
nitrogen. AllthenickeldiffractionpeaksofNi/silkfiberheatedin
nitrogen became much sharper than those of the original sample,
implying that the crystalline size of the sample was increased
during high temperature heat treatment in nitrogen.
Fig. 5 shows morphology of the Ni hollow fiber annealed at
220 C for 2 h in air. Compared Fig. 5 with Fig. 4b, it can be
seen that the surface of the Ni hollow fiber is much more porous
than that of the Ni/silk fiber, and there exists a large number of
◦
Fig. 4 shows SEM morphology of natural surface of pure
nickel powders and Ni/silk fiber prepared by electroless plating.
It is seen that these pure Ni particles possesses global struc-
tures with diameters of about 0.5–1.5 ␮m (Fig. 4a). However,
the particle is congregated together, which can be attributed to
the nickel’s magnetic attractive force. Fig. 4b shows the typical
morphology of the Ni/silk fiber by electroless plating. Compar-
ing Fig. 4b with Fig. 1 (the morphology of the starting silk)
reveals that the surface of Ni/silk fiber is much rougher than that
of the starting silk demonstrating that some new phases formed
on the surface of starting silk. Fig. 4c shows that the formed
Ni/silk fiber has several different shapes. Some fiber interweave
each other and form sphere-like structure (as shown in point 1,
Fig. 4c), some fiber form a spiral-like structure (point 2, Fig. 4c)
and some silk keep the starting original shape of the template
with a length of about 70 ␮m (point 3, Fig. 4c). Fig. 4d and
Table 1 shows that elements Ni, C, Si and O are found in Ni/silk
Table 3
The EDS results of Ni hollow fiber in Fig. 6a
Spectrum
Si
Ni
99.48
99.61
100.00
99.35
Total
Spectrum 1
Spectrum 2
Spectrum 3
Spectrum 4
Spectrum 5
0.52
0.39
0.00
0.65
0.74
100.00
100.00
100.00
100.00
100.00
99.26
Fig. 7. Frequency dependence of complex permittivity of ultrafine Ni/silk fiber
and Ni hollow fiber (heat treatment in nitrogen at 700 C for 1.5 h)–paraffin wax
Mean
0.46
99.54
100.00
◦
All results in weight percent.
composites.

H. Zhang, Y. Liu / Journal of Alloys and Compounds 458 (2008) 588–594
593
small pores on the surface of Ni hollow fiber due to oxidation
of the silk and removal of the CO and CO2 gas. The Ni hollow
fiber is composed of parallel nanowires, which are arrayed along
the axis of the starting silk. The nanowires also consisted of the
nickel particles with size of less than 100 nm, which is almost
the same as that of the Ni/silk fiber indicating that the nano-
the results demonstrate that Ni hollow fiber is prepared in this
work.
Fig. 7 shows the frequency dependence of complex permit-
◦
tivity of Ni/silk fiber and Ni hollow fiber (annealed at 700 C for
1.5 h in nitrogen) in 2–18 GHz. It indicates:
◦
ꢀ ꢀꢀ
(1) For Ni/silk fiber, ε and ε decreases with measuring fre-
sized Ni particles do not grow much at 220 C for 2 h. The EDS
ꢀ
ꢀ
results (Table 2) of the Ni hollow fiber in Fig. 5a shows that the
elements Ni, Si and O are found in the samples. The observed
silicon element is ascribed to the silicon plate bearing composite
powder, and the little amount of oxygen element may be due to
partialoxidationofcoatednickel. Theabsenceofcarbonelement
indicates that all the silk is removed during the heat treatment
process.
quency increasing, the value of ε at low frequency is as
high as 200, and it possesses a value of about 40 even at
18 GHz.
ꢀ
ꢀ
(2) For Ni hollow fiber, ε decreases with measuring frequency
ꢀ
ꢀ
increasing, the values of ε at low frequency are as high as
60, and it keeps a value of about 10 even at 18 GHz. The
ꢀ
value of ε increases with frequency up to about 4 GHz, after
ꢀ
that, it decreases; the value of ε at 18 GHz is about 10.
Fig. 6 exhibits particle morphology of the Ni hollow fiber
◦
ꢀ
ꢀꢀ
heated at 700 C for 1.5 h in nitrogen. The fracture section of the
(3) The values of ε and ε of Ni/silk fiber are much higher than
that of the Ni hollow fiber. The reason for this difference
needs further investigation.
fiber in Fig. 6a and b confirms that the hollow fiber is formed.
Comparing Fig. 6c and d with Fig. 5c and d reveals that no
nanowire was formed for the samples annealed in nitrogen, and
the particle size of nickel in the sample heated in nitrogen is
much larger than that of the sample annealed in air, which can
be ascribed to the much higher temperature used for the sample
fired in nitrogen. It is seen from Table 3 (the EDS results of the
fiber in Fig. 6a) that only the elements Ni and Si are found. All
ꢀ
Fig. 8 shows that μ had the value 1.0 more or less in all
measuring frequencies for both Ni/silk and Ni hollow fiber,
ꢀ
ꢀ
and no resonance peaks is observed in the μ spectrum. It
is due to the microwave magnetic loss of magnetic materials
originates mainly from domain wall resonance and natural ferro-
magnetic resonance. The domain wall resonance occurs only in
Fig. 8. Frequency dependence of complex permeability of ultrafine Ni/silk fiber
and Ni hollow fiber (heat treatment in nitrogen at 700 C for 1.5 h)–paraffin wax
Fig. 9. Frequency dependence of dielectric and magnetic loss of ultrafine Ni/silk
fiber and Ni hollow fiber (heat treatment in nitrogen at 700 C for 1.5 h)–paraffin
◦
◦
composites.
wax composites.

5
94
H. Zhang, Y. Liu / Journal of Alloys and Compounds 458 (2008) 588–594
multidomainmaterialsandusuallyinthe1–100 MHzrange. And
it has been reported that the natural resonance frequency for Ni
particles (dm = 1.4 ␮m) is about 1.8 GHz [27]. In this study, the
permeabilities were measured in frequency range of 2–18 GHz,
so neither domain wall resonance nor natural ferromagnetic res-
onance is observed. The value of μ is relatively low and the
maximum value is no more than 0.3 in all 2–18 GHz.
Fig. 9 shows the frequency dependence of dielectric loss
and magnetic loss in 2–18 GHz for Ni/silk fiber and Ni hollow
fiber–paraffin wax composites. It indicates that:
Acknowledgement
This work was financially supported by “The Science Fund
for Distinguished Young Scholars of Henan Province, China”
(Contract No. 0512002400).
ꢀꢀ
References
[
[
[
1] M.Z. Wu, H.H. He, Z.S. Zhao, X. Yao, J. Phys. D: Appl. Phys. 34 (2001)
069–1074.
2] M.Z. Wu, Z.S. Zhao, H.H. He, X. Yao, J. Magn. Magn. Mater. 217 (2000)
9–92.
3] M. Matsumoto, Y. Miyata, IEEE Trans. Magn. 33 (1997) 4459–4464.
1
8
(
1) For Ni/silk fiber, the dielectric loss is observed to be very
high with values of more than 4 in all frequencies. However,
the magnetic loss of Ni fiber–paraffin wax composites is low
with values of about 0.1–0.2 in 2–18 GHz. Two dielectric
loss peaks is observed in the range of 2–18 GHz, and the
corresponding frequency is about 8.96 and 14.48 GHz. The
maximum dielectric loss measured is up to 7.6.
[4] P. Chen, R.X. Wu, T.E. Zhao, F. Yang, Q. John, J. Phys. D: Appl. Phys. 38
2005) 2302–2305.
5] I.T.H. Chang, Z. Ren, Mater. Sci. Eng. A 375–377 (1/2 (special issue))
2004) 66–71.
(
[
[
(
6] Ph. Toneguzzo, G. Viau, O. Acher, F. Guillet, E. Bruneton, F. Fievet-
Vincent, F. Fievet, J. Mater. Sci. 35 (2000) 3767–3784.
[7] R.C. Che, Y.Q. Li, Z.H. Chen, H.J. Lin, J. Mater. Sci. Lett. 18 (1999)
963–1964.
1
(
(
2) For Ni hollow fiber, the dielectric loss decreases with fre-
quency increasing, and its value is observed to be high up
to 7 at 2 GHz and about 1 even at 18 GHz. However, the
magnetic loss of Ni hollow fiber is close to that of Ni/silk
fiber.
[
8] D.L. Griscom, J.J. Krebs, A. Perez, M. Treilleux, Nucl. Instrum. Methods
Phys. Res., Sect. B: Beam Interact. Mater. Atoms B32 (1988) 272–278.
9] R.D. Bruyne, Adv. Mater. Processes 147 (1995) 33.
[
[10] P. Rudkowski, G. Rudkowska, J. Appl. Phys. 69 (1991) 5017.
[11] P. Rudkowski, G. Rudkowska, A. Zaluska, J. Strom-Olsen, IEEE Trans.
Magn. 28 (1992) 1899.
3) The particular high dielectric loss of both Ni/silk and Ni
hollow fiber may be due to the anisotropy of the prepared
[
[
[
12] J. Strom-Olsen, Mater. Sci. Eng. A 178 (1994) 239.
13] G. Liu, Rare Met. Mater. Eng. 23 (1994) 7.
14] F.Z. Kong, X.B. Zhang, W.Q. Xiong, F. Liu, W.Z. Huang, Y.L. Sun, J.P.
Tu, X.W. Chen, Surf. Coat. Technol. 155 (2002) 33–36.
ꢀ
ꢀ
Ni fiber. The reason of the different behavior of (ε /ε)–f
curve of Ni/silk fiber and Ni hollow fiber may be due to the
obvious different crystalline size of those two kinds of fiber.
[15] H.P. Liu, G.A. Cheng, R.T. Zheng, Y. Zhao, J. Mol. Catal. A: Chem. 225
2005) 233–237.
(
[
[
[
16] Y.J. Chen, M.S. Cao, Q. Xu, J. Zhu, Surf. Coat. Technol. 172 (2003) 90–94.
17] Q.Y. Zhang, M. Wu, W. Zhao, Surf. Coat. Technol. 192 (2005) 213–219.
18] S. Shukla, S. Seal, Z. Rahaman, K. Scammon, Mater. Lett. 57 (2002)
4
. Conclusions
Ni hollow fiber was prepared by electroless plating and tem-
151–156.
plate removing process using hydrazine hydrate, NiSO4, a kind
[19] A.X. Zeng, W.H. Xiong, J. Xu, Mater. Lett. 59 (2005) 524–528.
◦
[
[
[
20] T. Kobayashi, J. Ishibashi, S. Mononobe, M. Ohtsu, H. Honma, J. Elec-
trochem. Soc. 147 (2000) 1046.
21] L.M. Ang, T.S.A. Hor, C.Q. Xu, C.H. Tung, S.P. Zhao, J.L.S. Wang, Carbon
of natural silk, etc. as starting materials. Heat treatment at 220 C
◦
for 2 h in air or at 700 C for 1.5 h in nitrogen atmosphere was
suitable for removing the starting silk template. The typical
length of the Ni hollow fiber was about 70 ␮m. The Ni hol-
low fiber heated in air was composed of nanowires parallel to
the axis of the starting silk. The dielectric loss of the prepared
Ni/silk fiber and Ni hollow fiber (heated in nitrogen) is high up
to 1 even at 18 GHz. The values of the magnetic loss of those
two kinds of fiber are about 0.1–0.2 in 2–18 GHz.
38 (2000) 363.
22] W.S. Chung, S.Y. Chang, S.J. Lin, Plat. Surf. Finish. 83 (1996) 68.
[23] S. Schaefers, L. Ras, A. Stanishevsky, Mater. Lett. 60 (2006) 706–709.
[
[
[
24] I.S. Kim, S.K. Lee, Scripta Mater. 52 (2005) 1045–1049.
25] W.B. Weir, Proc. IEEE 62 (1974) 33–36.
26] E.J. Vanzura, J.R. Baker-Jarvis, J.H. Grosvenor, IEEE Trans. Microwave
Theory Tech. 42 (1994) 2063–2070.
[27] G. Viau, F. Ravel, O. Acher, et al., J. Appl. Phys. 76 (1994) 6570–6572.