Вы находитесь на странице: 1из 5

Bonfring International Journal of Power Systems and Integrated Circuits, Vol.

1, Special Issue, December 2011 43


ISSN 2250 1088 | 2011 Bonfring
Abstract--- This paper presents a high-Q MWCNT (Multi
Wall Carbon Nano-Tube) network based pulse-shaped on-chip
embedded nano-inductor for power electronic circuits. This
high-Q inductor is fabricated using a composite of MWCNT
(multi Wall Carbon Nano Tube) network and Fe in
semiconductor processing environment. A layer of nano
composite film (Cu/CoFe2o4) is patterned as an introduction
layer of the nano-inductor. The performance of the fabricated
nano-inductor is compared with and without the nano
composite film. An inductance of 6.25 nH, with a quality
factor of 186 at 2.4 GHz is measured from the MWCNT
network inductor (with nano composite film). The maximum
inductance of the nano inductor is about 6.6 nH, and the
maximum Q factor is about 440. However the inductance
value and the quality factor is improved that of conventional
micro scale inductors, the chip area of the inductor is reduced
only 25%.
Keywords--- Nano-Inductor, power electronics, high Q.

I. INTRODUCTION
NDUCTORS are one of the bulk devices which limit the
performance in most of the power electronic circuits. During
19th century was Micro Electro Mechanical inductors played
an important role in replacing conventional copper inductors,
since the power consumption of the later is higher due to the
high current density. As an example the maximum current
density of a twenty turns of 16 AWG (American Wire Gauge)
inductor with 26 mm2 diameter will be about 275 A/cm2.
Micro-scale inductors have been investigated by several
researchers, and it has been proved that they can be superior to
conventional copper inductors [1-4]. However, these
investigations have also revealed that phenomena such as the
skin effect, proximity effects, eddy current losses, large power
consumption, and electric field penetration into the substrate
degrade the performance of on-chip inductors at high
frequencies. The primary challenge in designing on-chip
inductor with high quality factor is reducing the substrate
losses and series resistance of inductors [5]. Micro-Electro-
Mechanical (MEM) inductors, with the advanced
micromachining techniques, such as Bipolar and CMOS
processes, offer high ohmic resistance, and confine the
substrate losses. On the other hand, to reduce the series
resistance, post processed Cu inductors are suggested [6]. To
improve the quality factor of the on-chip inductor several
approaches have been proposed [7-10].
In the last decade of 19th century tubular structures of

B.S. Sreeja, SSN College of Engineering, Chennai.
S. Radha, SSN College of Engineering, Chennai.
CNTs are identified. There are two different types of CNTs
that can have high structural perfection. SWCNT consist of a
graphite sheet seamlessly wrapped into a cylindrical tube.
MWCNT comprise an array of such nanotubes that are
concentrically nested like rings. The unique electrical
properties of CNTs are to a large extent derived from their
peculiar electronic structure. CNT interconnect has a lower
skin effect compared to copper interconnects [11]. Moreover,
unlike copper interconnect, the resistance of the CNT
interconnect increases by a smaller amount or remains
unchanged at high frequencies [12]. In addition to magnetic
inductance, CNT offers kinetic inductance due to the smaller
momentum relaxation time. Moreover it is obvious form the
relation

that the magnetic field (H) induced by the current (I) in CNT is
about several thousand times greater than that induced by the
current in normal copper wire whose radius is several thousand
times greater than that of CNTs.
As in the relation

the magnetic field (H) is directly proportional to the inductance
(L), as large magnetic field results in large inductance. With
this properties, CNT interconnect is an appropriate material for
on-chip inductors for GHz applications. Implications of on-
chip inductor using CNTs had been investigated thoroughly
and the detailed performance analysis is presented in [13-14]. It
is subsequently shown that CNT based inductors offer three
times higher Q factor than their copper based counterparts,
without using any magnetic materials or Q factor enhancement
techniques.

Figure 1: SEM Images of CNT Structures (a) CNT Network
Structure (many small length CNTs scattered in an area) (b)
Densified CNT Bundle Array (c) CNT Bundles ( long CNTs
WHICH are Continuous and Intact) (d) Well Aligned CNT
bundles

Fabrication and Characterization of High-Q Nano-
Inductor for Power Electronics
B.S. Sreeja and S. Radha
I
Bonfring International Journal of Power Systems and Integrated Circuits, Vol. 1, Special Issue, December 2011 44
ISSN 2250 1088 | 2011 Bonfring

(a) (b)

(c)
Figure 2: (a) CNT Network without Conductive Paths (b) CNT
Network with Conductive Paths (c) Schematic illustration of
MWCNT based Pulse-Shaped Embedded Nano-Inductor with
its Dimensions

(a) (b)

(c)
Figure 3: Characteristics of the Nano Composite Inductor
Plotted as a Function of frequency (a)Relative permittivity of
MWCNT network (70% MWCNT+ 30%Fe) (b) Relative
Permeability of MWCNT Network (70% MWCNT+ 30%Fe)
(c) Sheet Conductances of MWCNT Networks with same
amount of Fe Catalyst where the Diameter of the MWCNT is
about 15~20 nm with an Average Length of 5 m.


II. BACKGROUND
A. CNT Networks:
CNT is a rolled graphite sheet, as example of a quasi-one
dimensional conductor have attracted considerable attention
since its discovery, and its structures are formed as bundles
(fig.1.(b), (c) and (d)) and networks (fig.1.(a)) as shown in
fig.1. Well aligned CNT bundles continue to receive attention
in electronics world, because of its attracting electrical and
magnetic properties. Moreover properties of properly aligned
CNT bundles are rigorously investigated in the literatures [15-
16]. The development of CNT bundles is limited by the lack
of effective bulk methods for precisely manipulating and
aligning nanotubes at the very fine nano scale. By coupling
magnetically susceptible components such as ferromagnetic
material with CNTs, CNT networks (CNTNs) can be created
and they provide easier manufacturability and cost effective
solutions. A random CNTN, as a semiconducting material for
Thin Film Transistors (TFTs), shows higher performance in
comparison with its organic counterpart, allows low-cost
fabrication, proves the possibilities of growth in future CNTN
applications [17]. The synthesis and deposition of CNTNs can
be accomplished using simple process steps without additional
complicated processes. Random CNTNs are formed by
directly growing CNTs on previously dispersed catalyst
islands with chemical vapour deposition (CVD) at high
temperatures.
III. PROPOSED INDUCTOR SCHEMATICS
The proposed inductor schematic is illustrated in Fig.2(c).
The effective volume of the inductor is approximately 0.09 x
.036 x .025 mm3. Each segment of the MWCNT network based
inductor can be represented by series inductances (kinetic, self,
and mutual), series resistances (contact, scattering, interface,
and sheet), parallel capacitance (electrostatic and parasitic) and
substrate parasitics.
Generally nanotubes are treated as good conductors due to
its high aspect ratio (L/D). Since the diameter (D) of the
nanotube is relatively very small compared to the length (L) of
the tube it exhibits good electrical conductivity. According to
this, the long MWCNT bundle will have more electrical
conductivity. In this work MWCNT networks (number of small
nanotubes embedded in a line), are used to form inductor lines.
MWCNT network with randomly distributed tubes will have
very high resistance and low conductivity which further
increases the resistive loss. Hence it is very important to keep
an eye on the continuity of the nanotube arrangement while
fabrication to improve the electrical conductivity of the
MWCNT network line.
Fig.2.(a) shows the structure of MWCNT network, the
electrical conductivity of the network is very low due to the
low density of tubes. With the increase of randomly distributed
nanotubes the distance among the nanotubes become smaller
gradually and complete electrical conducting paths to transfer
electric charges has been formed as shown in Fig.2.(b). As this
stage, the electrical conductivity increases remarkably. Further
increasing the density increases conductivity, but finally tend to
be constant at high volume fractions of CNTs. The effective
Bonfring International Journal of Power Systems and Integrated Circuits, Vol. 1, Special Issue, December 2011 45
ISSN 2250 1088 | 2011 Bonfring
conductivity of the embedded line depends on the length,
diameter, concentration and interface properties of the
MWCNTs. The effective conductivity of the inductor is shown
in Fig. 3(c). The conductivity increases with MWCNT network
density. At high frequencies the conductivity remains almost
constant.
IV. DEVICE FABRICATION
The proposed inductor was fabricated on a glass substrate.
A glass substrate was chosen to prevent degradation of self
resonant frequency at high frequencies. A SU-8 mold was used
to pattern the inductor lines on the glass substrate. SU-8 mold
does not show aggression with the organic chemicals and rigid
for insulating layer. The thickness of the patterned layer was
around 10 m. 5 nm/120 nm/5 nm
Ti(200A)/Cu(2000A)/Ti(200A) seed layer were deposited by
RF sputter on the patterned substrate. The most top layer is
utilized as an adhesion promotion layer for the PR molding
process. Super-paramagnetic CoFe2O4 nanocomposite powder
with the size of less than 55 nm is chosen to prepare the
solution. This nano composite solution was prepared in an
alkaline non-cyanide copper plating solution to avoid a
chemical reaction because the conventional acid sulfate copper
plating solution reacts with the CoFe2O4 nanocomposite and
causes very little CoFe2O4 nanopowder incorporation in Cu
matrix. A layer of Cu/CoFe2O4 nano composite was developed
inside the mold.

Figure 4: Micrograph of MWCNT Network based
Embedded Inductor
Frequency
(GHz)
Inductance (nH) Q factor
0.1 0.6 50
1 4.2 150
2 6.6 160
3 6.0 200
4 4.3 372
5 4 440
6 3 420
7 2 400
8 1.5 290
9 1 210
10 0.9 205


CNT droplet was made by dispersing MWCNT (CVD
MWNT 95) powder (70%) and Fe (30%) in ethanol. In this
work the average length of the CVD MWNT 95 was ~5 m
with an average diameter of 15~20 nm. The aspect ratio of the
MWCNTs is 250 approximately. Fe is used as a catalyst which
results in Fe being incorporated into the CNTs. Fe is an
impurity magnetic material and is purposely used in order to
produce an inductive phase at low frequency [7]. Additionally
CNT-Fe nanoparticles are much stronger and have higher
conductivity than conventional and classical particles. A well-
dispersed CNT droplet was deposited over the nanocomposite
film and dried using evaporation. Ethanol was evaporated by
using the vacuum chamber.
After deposition, the developed structure was patterned by
photolithography and the residual MWCNT deposition on the
mold was removed by rubbing. The filling and rubbing process
was repeated few times and deionised water was dropped to get
high packing density of MWCNT networks in the inductor
lines. Pressing and stretching the CNTs can change their
electrical properties by changing the quantum states of the
electrons in the carbon bonds. MWCNT inductor was
encapsulated by Cu since the quality factors can be improved
by Cu electroplating. Cu electroplating in another way covers
the gap between the MWCNTs. The maximum current density
was 12 mA/ cm2. The structure was fabricated with and
without the nanocomposite film to verify the effects caused by
the film. Fig. 4 shows an SEM image of the fabricated CNT
network based inductor.
A. Measurement Results and Discussion
The s-parameters were measured with the help of an
HP8510C network analyzer and a corresponding cascade
Micro-tech ground-signal-ground probing station in the
frequency range of 0.1 GHz to 10 GHz. Dummy pads were
designed to extract the parameters of the designed inductor.
The parasitic parallel capacitance and the series contact
resistance between the substrate and the contact pads of the
inductor are de-embedded via the dummy patterns and the
measured S-parameters are then transformed to Y-parameters
from which the inductance and the quality factors are
calculated using equation (1).

B. Electrical Characterization
The high frequency permittivity and permeability
characteristics were estimated from the S parameters using
vector network analyzer. Two port measurements were
employed, the measured permeability varies from 5 to 15(refer
Fig 3.(b)) depending on the ferrite loading and the tangent
losses were less than 0.001 dB when the ferrite 4 loading was
above 25%. The permittivity of the MWCNT network was
measured between 7.5 to 18 over a frequency range of 0 to 10
GHz as shown in Fig.3.(b).
Bonfring International Journal of Power Systems and Integrated Circuits, Vol. 1, Special Issue, December 2011 46
ISSN 2250 1088 | 2011 Bonfring
C. Q-Factor of nano inductor
Quality factor, an extremely important figure of merit of the
inductor is the ratio of energy stored to energy dissipated.

where E
stored/cycle
is the energy stored in the inductor per cycle,
while E
lost/cycle
is the energy lost in the inductor per cycle. Pdiss
represents the average power dissipated by the inductor, and
E
mag
and E
elect
represent the averages of the stored magnetic
(due to inductance) and electric energies (due to capacitance),
respectively. Q factor can also be regarded as a measure of the
ratio of the wanted quantity, related to the inductive reactance
to the unwanted quantity (resistance). At low frequencies
(capacitive effect is low), the Q factor increases linearly with
frequency. As frequency increases, capacitive and magnetic
coupling with substrate become more and more significant,
which increases energy dissipation. The capacitive loss
becomes significant near the self resonant frequency (SRF).
The SRF of the MWCNT inductor is very high (>>20 GHz),
beyond the range of the network analyzer, hence the capacitive
loss is small and negligible.

Figure 5: Measurement Results of the MWCNT Inductor
showing Inductance and Q Factor with Respect to Frequency

Figure 6: Hysteresis characteristics of Magnetic Field Density
(H)
In addition, the skin/proximity effects of the conductor also
increase the loss (eddy current loss) of the inductor. In this
work the nanocomposite film on the substrate highly reduces
the skin effect due to its high resistivity. However the eddy
current loss is minimized by the nanocomposite film, a 10%
degradation in the Q-factor is measured at the low frequencies
(0.1 GHz-1.2 GHz) in comparison with that of the MWCNT
inductor without nanocomposite film. It may be resulted by the
increase of the electrical resistivity. Together with low skin
effect and attracting properties of nano tubes the inter leg
capacitance effects were comparatively small. with More the
non-conductive second phase particles incorporated into a
metal matrix, the higher the electrical resistivity of the film is
measured. However CNT offers large kinetic inductance, the
impact of kinetic inductance on the Q-factor improvement is
negligible. To produce high Q inductors the magnetic-field
induced inductance should be maximized. Since the magnetic
inductance is dependent on the geometry and area, it can be
expected that CNT-based high Q inductors would require
similar area as conventional metal inductors. The addition of
superparamagnetic nanoparticles in to the Cu matrix increases
the Q factor at high frequencies via the high frequency
characteristics (> 1 GHz) of the superparamagnetic nano
particles. Fig.6. shows the SQUID measurement results at
room temperature. Uniform magnetic flux pattern is seen. The
hysteresis characteristic of the device shows the ferromagnetic
properties. Without largely sacrificing the reduction of
electrical conductivity, the fabricated nano inductor shows its
potential for the fabrication of high performance cost effective
high Q on chip inductor due to its effective reductions of size,
and eddy current loss. Given that the inductance is 6.3 nH and
total resistance of the inductor is extracted as a 530 m, the
quality factor is approximately 179 at 2.4 GHz. Fig 5 shows the
measurement results of the inductor.
V. CONCLUSION
In this paper we proposed a pulse structure for MWCNT
bundle inductor. This CNT network based inductor is
fabricated on a glass substrate. The device is tested with and
without nano composite film. It is noted from the performance
comparison, that the addition of nano composite layer enhances
the Q factor. At 2.4 GHz, the commercially used frequency
range, we measured a Q factor of 226 and an inductance of
28.3 nH. The fabricated device produces a maximum Q factor
of at 6.2 GHz. The self resonant frequencies are above 20 GHz,
which are above the range of the network analyzer used. The
self resonant frequencies are predicted to be in the range of 100
GHz-150 GHz, and with these self resonant frequencies the
proposed inductor can be used for power electronic
applications since the energy loss is low. Further, the analysis
in extracting the equivalent circuit parameters is expected. Also
with further improvements in the fabrication the inductance
could be increased at high frequencies above 5 GHz.
REFERENCES
[1] H. Sugawara, Y. Yoshihara, H. Ito, K. Okada, and K. Masu, Wide-
range RF variable inductor on Si CMOS chip with MEMS actuator, in
Proc. 34th Eur. Microw. Conf., Oct. 2004, vol. 2, pp. 701704.
[2] J. Salvia, J. A. Bain, and C. P. Yue, Tunable on-chip inductors up to 5
GHz using patterned Permalloy laminations, in IEDM Tech. Dig., Dec.
2005, pp. 943946.
[3] I. Zine-El-Abidine, M. Okoniewski, and J. G. McRory, RF MEMs
tunable inductor using bimorph microactuators, in Proc. Int. Conf.
MEMS, Nano Smart Syst., Jul. 2005, pp. 436437.
[4] C. P. Yue and S. S. Wong, Physical modeling of spiral inductor on
silicon, IEEE Trans. Electron Devices, vol. 47, no. 3, pp. 560568,
Mar. 2000.
[5] J.Cranickx and M.Steyaert, Wireless CMOS Frequency Synthesizer
Design. Norwell, MA: Kluwer, 1998.
Bonfring International Journal of Power Systems and Integrated Circuits, Vol. 1, Special Issue, December 2011 47
ISSN 2250 1088 | 2011 Bonfring
[6] H. Hsieh and L.-H. Lu, A high-performance CMOS voltage-controlled
oscillator for ultra-low-voltage operations, IEEE Trans. Microw.Theory
Techniques, vol. 55, no. 3, pp. 467473, Mar. 2007.
[7] Hsiao-Chin Chen, Chao-Heng Chien, Hung-Wei Chiu, Shey-Shi Lu,
Kung-Neng Chang, Kun-Yu Chen, and Shi-Hao Chen, A Low-power
Low-phase-noise LC VCO with MEMS Cu Inductors, IEEE Microwave
and Wireless Components Letters, vol 15, no 6, pp.434-436, Jun. 2005.
[8] E.-C. Park, Y.-S. Choi, J.-B. Yoon, S. Hong, and E. Yoon, "Fully
integrated low phase-noise VCOs with on-chip MEMS inductors", IEEE
Transactions on Microwave Theory and Techniques, vol. 51, no. 1 (Part
2), pp. 289-296, Jan. 2003.
[9] P.Park, C. S. Kim, M. Y. Park, S. Do Kim, and H. K. Yu, Variable
inductance multilayer inductor with MOSFET switch control, IEEE
Electron Device Lett., vol. 25, no. 3, pp. 144146, Mar. 2004.
[10] S. Zhou, X. Q. Sun, and W. N. Carr, A micro variable inductor chip
using MEMS relays, in Proc. Int. Conf. Solid-State Sensors
Actuators, Jun. 1997, vol. 2, pp. 11371140.
[11] H.-C. Chen, C.-H. Chien, H.-W. Chiu, S.-S. Lu, K.-N. Chang, K.-Y.
Chen, and S.-H.A. Nieuwoudt and Y. Massoud, Predicting the
Performance of Low-Loss On-Chip Inductors Realized Using Carbon
Nanotube Bundles, IEEE Transactions on Electron Devices, vol. 55,
no. 1, pp. 648651, January
[12] N. Srivastava, H. Li, F. Kreupl, and K. Banerjee, On the applicability
of single-walled carbon nanotubes as VLSI interconnects, IEEE
Trans.Nanotechnol., vol. 8, no. 4, pp. 542559, Jul. 2009.
[13] H. Li, and K. Banerjee, Carbon Nanomaterials for Next-Generation
Interconnects and Passives: Physics, Status, and Prospects IEEE
Trans.on Electron devices, vol. 9,no.9, 1799-1821, September 2009.
[14] H. Li and K. Banerjee, High-frequency effects in carbon nanotube
interconnects and implications for on-chip inductor design, in IEDM
Tech. Dig., 2008, pp. 525528.
[15] Y. Massoud and A. Nieuwoudt, Modeling and Design Challenges and
Solutions for Carbon Nanotube-Based Interconnect in Future High
Performance Integrated Circuits, ACM Journal on Emerging
Technologies in Computing Systems, vol. 2, no. 3, pp. 155196, July
2006.
[16] Miguel A. Correa-Duarte,Nicholas Wagner,,Jos Rojas-hapana,Christian
Morsczeck,Michael Thie, and and Michael Giersig, Fabrication and
Biocompatibility of Carbon Nanotube-Based 3D Networks as Scaffolds
for Cell Seeding and Growth, Nano Letters 2004 4 (11), 2233-2236
[17] Jae Won Jang, Kyu Won Lee, Cheol Eui Lee, Tae Jae Lee2, Cheol Jin
Lee and Seung Chul Lyu, Magnetic properties of Fe catalysts included
in carbon nanotubes phys. stat. sol. (b) 241, No. 7, 16051608
(2004)