Вы находитесь на странице: 1из 7

IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO.

11, NOVEMBER 2008

4089

Fabrication and Analysis of High-Performance Integrated Solenoid


Inductor With Magnetic Core
Dok Won Lee1 , Kyu-Pyung Hwang2 , and Shan X. Wang1;3
Department of Materials Science and Engineering, Stanford University, Stanford, CA 94305 USA
Intel Corporation, Folsom, CA 95630-9501 USA
Department of Electrical Engineering, Stanford University, Stanford, CA 94305 USA
Integrated solenoid inductors with magnetic core were fabricated and analyzed. An inductance above 70 nH was achieved while
keeping the coil resistance below 1 and the device area below 1 mm2 using a solenoid design with a single magnetic layer. The inductance of the magnetic inductor was more than 30 times that of the air core inductor of the identical geometry, and the quality factor of
the magnetic inductor was 5. Novel inductor designs and the scalability were also examined, and an inductance density higher than
200 nH/mm2 was obtained. The measured device properties and engineering tradeoffs were well explained by analytical models we
developed.

Index TermsAmorphous magnetic film, magnetic losses, permeability, power delivery applications, thin film inductor.

I. INTRODUCTION

OWADAYS cell phones and laptop computers are playing


important roles in our everyday lives, and the demand for
more portable electronic devices continues to increase rapidly.
This is currently driving the integration or embedding of passive components, which would replace off-chip discrete modular assemblies. Meanwhile, various approaches of integration
in system-on-a-chip (SoC) and system-in-package (SiP) circuits
are gaining much attention. However, the poor properties of integrated inductors have been a critical factor limiting the overall
performance of radio-frequency (RF) circuits and hence the realization of RF SoC or SiP circuits. The current mainstream integrated inductor is air core spiral inductor, which typically has
a small inductance while consuming a relatively large area [1].
In addition, due to the large coil resistance, the quality factor
becomes significant only in GHz frequency range.
Use of magnetic core with high permeability in the integrated
inductor was proposed decades ago to significantly increase the
of the magnetic mainductance by the relative permeability
terial used [2]. This, in turn, can significantly reduce the device
area and the coil resistance needed to meet the inductance requirement. However, the inductance gain by over the air core
inductor has not been realized in experiments, and the reported
inductance enhancements so far have been typically very limited
[3], [4]. Recently there have been several reports of the large enhancement in inductance with the use of magnetic material [5],
[6]; however, the inductance enhancement was still significantly
smaller than of the magnetic material used. While there have
been a number of qualitative discussions and quantitative analysis to understand the realizable inductance [2], [7], [8], the discrepancy still needs to be carefully examined. In addition, the
use of magnetic core comes with the cost of introducing magnetic power losses. This needs to be well understood in order to
make the magnetic inductor practical and useful.
In this study, we present the high performance integrated inductors using a solenoid design with a single magnetic layer.

Digital Object Identifier 10.1109/TMAG.2008.2003398

Fig. 1. Schematic design of an integrated solenoid inductor: (a) top view and
(b) cross-section view.

The inductor devices show very high inductance enhancement


over the air core equivalent and large inductance density while
maintaining low coil resistance and small area consumption.
Furthermore, the device properties can be accurately described
by the analytical model we developed and used in the design
process. Our findings indicate that the integrated magnetic inductors can now be reliably designed and fabricated for various
applications in the RF circuit and mobile power delivery, enabling the realization of the RF integrated electronics.
II. INDUCTOR DESIGN AND FABRICATION
A. Analytical Models
Various designs for the magnetic inductors have been reported in the literature. Among them, the solenoid inductor
design is considered in this study, because it resembles the
discrete wire-wound inductor and is very efficient magnetically
[9]. A schematic design of an integrated solenoid inductor
with magnetic core is shown in Fig. 1. Design parameters used
in the models are as follows: the number of turns , width
, length of air core
or magnetic core
,
of air core
, thickness
thickness of air core , width of magnetic core
, widths of coil
or
, length of coil
of magnetic core
, thickness of coil , gap between turns , size of via
,
and gap surrounding the via .

0018-9464/$25.00 2008 IEEE

4090

IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 11, NOVEMBER 2008

We put forward the following analytical models for the device properties based on the established analytical models for
discrete solenoid inductors [10], the careful analyses of experimental results [3], [4], and the comparison with the results from
the finite element electromagnetic field simulation tool, Ansoft
HFSS [11]. We have confirmed previously that it has the ability
to describe the properties of the actual integrated inductor devices [4], [12].
and coil resistance
of the integrated
The inductance
solenoid inductor with air core are as follows [9]:

(1)
(2)
is the winding inductance,
is the parwhere
asitic inductance, and is the electrical resistivity of the coil
material. The expression for the inductance is modified from
the classical Wheeler formula based on the comparison with the
depends on the
experimental and simulation results.
, which
modified cross-section area of the air core,
completely includes the vias and coils surrounding the air core.
represents the effects of deviations from the classical
winding, including the probe pads at the ends of winding and the
ground ring surrounding the inductor. The expression for the resistance is obtained by representing it as the series resistance of
three line segments: the coil crossing over the air core, the connection to the via, and the via. The contact resistance between
two conductor layers is very small in well fabricated devices and
thus negligible.
The classical expression given by Soohoo has been widely
used to estimate the inductance of magnetic inductors. However, it was found to overestimate the actual inductance significantly [12]. One of the reasons for the discrepancy is the demagnetization effect. The presence of the demagnetization field
formed inside the finite-sized magnetic core effectively reduces
of the magnetic core. In addition,
the relative permeability
the contribution from winding to the inductance can be comparable to the magnetic contribution for integrated inductors, and
hence the winding and magnetic contributions should be carefully separated. Taking these into account, we describe the inductance of the integrated solenoid inductor with magnetic core
by the following expression [12]:

The inductance enhancement due to the magnetic contribuand


tion can be estimated using (1) and (3). Assuming
, the inductance enhancement or gain can
be expressed as

(4)
is the cross-section area of the magnetic core. The
where
above expression shows that the actual inductance gain is greatly
and
. The inreduced from , since
ductance gain can however still be significant. In this study, the
enhancement was by more than a factor of 30, as will be shown
later.
Use of magnetic core comes with the cost of introducing
magnetic power losses and results in a significant increase in
the inductor resistance at higher frequencies than at DC. Main
magnetic loss mechanisms include the hysteresis loss, the eddy
current loss, and the ferromagnetic resonance loss [14], [15].
Based on the classical electromagnetism, the energy stored due
, and the magto the inclusion of the magnetic core,
are related to the real and imaginary
netic power loss
parts of the permeability of the magnetic material used, respectively [16]:
(5)
(6)
Since the real and imaginary parts of the permeability are significant inside the magnetic core and assuming that they are unican be related to the
form inside the magnetic core,
and
as follows:
permeability ratio
(7)
Equation (7) can be expressed in terms of device properties
using the following relations:
(8)

(9)
where
is the contribution to the resistance due to the
is the total energy stored in the
magnetic power losses,
is the energy stored due to the
magnetic inductor, and
coils of the inductor. Substituting (8) and (9) into (7), we obtain
the resistance of the integrated solenoid inductor with magnetic
as follows:
core
(10)

(3)
is the net increase in the inductance due to the magwhere
is the demagnetization factor. The
netic contribution and
demagnetization field is not uniform inside the magnetic core
having the orthorhombic shape, and it is not trivial to estimate
analytically for
[13]. However, numerical solutions
are available in the literature, and it is shown the higher the permeability, the more severe the demagnetization effect [13].

As the frequency increases,


becomes much more sig, because it is proportional to both the frenificant than
. The permeability ratio itself also increases
quency and
instead of
rapidly with the frequency. By using the ratio
to express the contribution due to the magnetic power losses,
its dependence on the structure and geometries of the device can
be represented by the inductance enhancement. The magnetic
core contribution to the resistance is then proportional to the net

LEE et al.: FABRICATION AND ANALYSIS OF HIGH-PERFORMANCE INTEGRATED SOLENOID INDUCTOR WITH MAGNETIC CORE

4091

inductance enhancement as shown in (10), which imposes a fundamental engineering tradeoff in magnetic inductor design: The
more inductance enhancement we obtain by using a magnetic
core at low frequencies, the more resistive losses we introduce
at high frequencies.
is the permeability ratio
It should be emphasized that
of the actual magnetic core used in the integrated inductor. The
permeability of the patterned and processed magnetic core is
usually not the same as that of the bulk magnetic film as shown
later.
is the figure of merit that determines the
Quality factor
efficiency of the inductor device. is proportional to the ratio
of energy stored to the energy lost, per unit time, and it can be
expressed as follows [10]:
(11)
Using (3) and (10), we obtain the quality factor of the magnetic
to be
inductor
(12)
can be significantly larger than the
At low frequencies
,
quality factor of the air core inductor,
, but it starts
because of the large inductance enhancement
due to the larger resistive losses at
to decrease sooner than
higher frequencies.
is very small,
becomes close to
according
If
to (12). This is intuitively correct, since a magnetic inductor
would behave like an air core inductor for a very small magnetic
is very large compared
contribution. On the other hand, if
,
approaches the permeability ratio
of the
to
increases with frequency while
magnetic core. In general,
decreases, and at the cross-over frequency, both
the ratio
is
quantities become equal. Equation (12) suggests that
below this cross-over frequency, and
behigher than
beyond the cross-over frequency. Hence,
comes less than
the cross-over point can be considered as the useful bandwidth
of the magnetic inductor, beyond which it is better to use air core
inductor of the identical geometry.
One should note that the magnetic core losses were mainly
considered in this study in deriving the frequency-dependent resistance and the quality factor of the magnetic inductor. Howresonance effect
ever, other loss mechanisms including the
should be also taken into account if that the parasitic capacitance is not negligible or that the operating frequency further
increases.
B. Inductor Design
A solenoid inductor was mainly considered here, in which a
magnetic bar was wrapped around with a coil of various numbers of turns. Design parameters were optimized using the above
equations to achieve a high inductance while maintaining the
and the coil resistance
.
lateral device area
Considering the practical design rules for the fabrication process
available and using the copper and the amorphous
(atomic %) alloy
[12] as the conductor and the magnetic core materials, respectively, we used the following set of
,
the parameters for the inductor fabrication:
,
,
,
,

Fig. 2. Schematic illustrations of the process steps for the integrated magnetic
inductors.

,
, and
. CoTaZr alloy
is chosen because of its high relative permeability to enhance
the inductance significantly and high resistivity
to reduce the eddy current loss.
In addition to the above solenoid inductor with a magnetic
bar at its core, which we sometimes refer to it as standard
solenoid, a design with closed magnetic core was considered
in attempt to improve the closure of magnetic flux and hence
to further increase the inductance. Two magnetic bars were put
parallel to each other, with the coils connected in series. The
magnetic bars were closed and formed a honeycomb-like shape,
which will be discussed further in Section II-C.
In practice, smaller design parameters would be desirable to
reduce the device size. Moreover, if only the lateral parameters
are reduced while keeping the vertical parameters unchanged,
the inductance is affected only by the change in the demagnetization factor without changing the coil resistance significantly
according to (2) and (3). In order to examine the effect of scaling
down, a set of the standard solenoid inductors was prepared with
the lateral parameters scaled down by a factor of two with the
vertical parameters unchanged.
C. Inductor Fabrication
The integrated inductors were fabricated on Si wafers with
to reduce the substrate loss and 1
resistivity
thermal oxide deposited. Fig. 2 illustrates the process
steps. Bottom copper conductors were formed by electroplating
selectively through the photoresist mask [Fig. 2(a)]. Cr/Cu
layers were used as the barrier/seed layers for the electroplating
process. Polyimide PI-2611 was spin-coated and hard-cured
to form the insulating layer [Fig. 2(b)]. A partial
at 350
planarization was achieved by the polyimide coating process,
but a better surface planarization is needed before the deposition of the magnetic layer in order to maintain the desirable
magnetic properties [17]. So the photoresist etch-back process

4092

IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 11, NOVEMBER 2008

Fig. 3. SEM images of the integrated magnetic inductors: (a) standard solenoid
; (b) solenoid after scaling down by a factor of two (scaling
with
) with
; (c) two standard solenoids having the coils connected in
series (series) with
for each magnetic core; and (d) series solenoid
inductor having the closed magnetic core (closed core) with
.

N = 8:5
N = 8:5

N = 8:5

Fig. 4. FIB cross-section images of the standard solenoid inductor with magnetic core, deposited on planarized polyimide.

N = 8:5

was applied to improve the planarization of the polyimide


layer [Fig. 2(c)]. Photoresist was blanket-coated or formed
selectively on top of the Cu conductor patterns, and then it was
plasma. In addition, dummy patterns
blanket-etched using
were added in the bottom conductor layer to help planarize
the polyimide coating. The step height caused by the bottom
by a proconductor layer was measured to be below 0.5
filometer after the planarization process. Amorphous CoTaZr
film was sputter-deposited and wet-etched using nitric acid to
form the magnetic core [Fig. 2(d)]. Polyimide was again coated
and hard-cured to form the insulating layer [Fig. 2(e)], and vias
were then formed by dry-etching the polyimide layer using
plasma [Fig. 2(f)]. Sputtered
film was used as the mask
layer for the via formation process. Top Cu conductors were
formed again by electroplating through the photoresist mask
[Fig. 2(g)]. Sputter-etching was carried out prior to the Cr/Cu
layer deposition to obtain the good contact between the top and
bottom conductors.
Fig. 3 shows the scanning electron microscope (SEM) images of the fabricated magnetic inductors with various designs.
Focused ion beam (FIB) cross-section images shown in Fig. 4
confirm the successful fabrication of multilayered inductor devices with the top and bottom conductor thicknesses of 6.6 and
, respectively. The successful polyimide planarization is
4.4
also confirmed, resulting in the continuous magnetic core layer
. The width of the magnetic core
with the thickness of 2.2
due to undercuts resulting from wet-etching.
was about 340
The permeability spectra are important for understanding the
effect of the magnetic power losses. In order to obtain the perthick Comeability spectra of the actual magnetic core, 2
TaZr magnetic core structures were processed in parallel with
the inductor fabrication. The frequency responses of the permeability were measured by a Ryowa permeameter using a pick-up
loop method [18]. The permeability spectra and the magnetic
domain pattern in Fig. 5 confirm that the magnetic core in the
fabricated inductor maintains the desired soft magnetic properties. The permeability spectra of blanket film and processed

m

Fig. 5. (a) Permeability spectra of 2


thick CoTaZr blanket film and processed magnetic core and (b) Kerr microscope image taken from the processed
magnetic core.

magnetic core structures are not identical to each other mainly


due to the demagnetization effect and the dependence on the
substrate surface [12].
III. RESULTS AND DISCUSSION
The device properties of the fabricated inductors were measured using an HP 8753ES network analyzer with two standard model 40A-GSG probes (GGB industries Inc.). The values
of the inductance, resistance and quality factor were calculated
from the Y-parameters of the inductor devices according to the
two-port model [19], [20]. Open and short de-embedding structures were used to de-embed the parasitic effects of the probe
pads and ground ring.
Fig. 6 shows the measurement data of the standard solenoid
inductors with different numbers of turns. The inductance increases with the number of turns and has a value of 70.2 nH at
turns. This is an enhancement by a
10 MHz with
factor of 34 from 2.0 nH of the air core inductor with the identical geometry. It is higher than the reportedly record-high enhancement of
[6], and it should be noted that a single
magnetic layer is used while [6] required two layers of magnetic materials and the formation of magnetic vias. The device
(not shown in Fig. 3) is 0.88
, correarea for
, and the DC
sponding to an inductance density of 80
resistance for
is 0.67 . The observed inductance

LEE et al.: FABRICATION AND ANALYSIS OF HIGH-PERFORMANCE INTEGRATED SOLENOID INDUCTOR WITH MAGNETIC CORE

Fig. 6. Measurement data of the standard solenoid inductors: (a) inductance


and (b) resistance.

Fig. 7. Comparison between the measurement and the calculation data for the
standard solenoid inductors: (a) inductance and (b) resistance.

density is higher than the typically reported


for coil resistances less than 1 [21]. However, it should be
noted that the device properties including the inductance enhancement and density are dependent on the inductor design
and the magnetic material used. The resistance increases significantly with the frequency, and the increase is greater for higher
number of turns. This can be well explained using (10). As the
frequency increases, the magnetic core contribution to the resistance becomes more significant than the coil contribution, and
it is greater for a larger inductance gain.
The ripples observed in the frequency-dependent device
properties in Fig. 6 are closely related to the intrinsic properties
in the permeability spectra of CoTaZr magnetic core and were
discussed previously [22].
The measured inductance values and the calculation results
using (3) were compared in Fig. 7(a). Numerical solutions from
[13] were used to estimate the demagnetization factors. Good
agreement confirms that the demagnetization effect plays a
major role in determining the effective permeability. While the
magnetic domain structure also needs to be considered in order
to fully account for the effective permeability of the magnetic
cores under various geometries [12], this indicates that (3) can
adequately estimate the inductance of the integrated magnetic
inductors. The inductance values of the air core inductors
also show good agreement with (1). The expected inductance
enhancement is about 30 according to (1) and (3), which is
very close to the observed enhancement of 34 . This is also
in good agreement with the enhancement of 30 predicted
by HFSS simulations. Low-frequency resistance data of the
air core and magnetic inductors agree well with the expected
coil resistances estimated by (2), confirming that the magnetic
integrated inductors were fabricated successfully [Fig. 7(b)].

4093

Fig. 8. Comparison of measurement and HFSS simulation data between different designs of the magnetic inductors: (a) inductance and (b) inductance
density.

Fig. 9. Measurement data of the standard solenoid inductors after scaling down
by a factor of two (scaling
): (a) inductance and (b) resistance.

=2

Fig. 8(a) compares the measured inductance values of various


inductor designs. The inductance values are almost twice for the
series solenoid inductors [series shown in Fig. 3(c)] in comparison to those for the standard inductors. The inductors with
closed magnetic core [closed core shown in Fig. 3(d)] have
further increased inductances, reaching as high as 158 nH with
at 10 MHz. However, the device area is more than
twice that of the corresponding standard inductor, and hence
the resulting inductance density is actually reduced as shown
in Fig. 8(b). This suggests that the closed magnetic core is not
effective in improving the magnetic flux closure significantly,
and it can be explained by the tensor nature of permeability of
the magnetic core [4]. This is confirmed by the fact that the
measurement data are in good agreement with the HFSS simulation results. We have confirmed previously that the software
has the ability to describe the tensor nature of permeability [4].
Hence, due to the uniaxial anisotropy and tensor nature of permeability, the effective shape of the honeycomb-shaped magnetic core should be viewed as two parallel magnetic bars closed
by two bad soft magnets.
While the novel inductor design did not improve the inductance density, the standard solenoid inductor is shown to be very
efficient, and its inductance density can be further increased
with scaling down. Fig. 9 shows the measurement data of the
magnetic solenoid inductors with the lateral parameters reduced
shown in Fig. 3(b)] from
by a factor of two [scaling
those of the standard solenoids. For
, the inductance
is 48.4 nH at 10 MHz, which is more than a half of 70.2 nH for
the equivalent standard inductor; however, the device area is re, resulting in the increase
duced by a factor of four to 0.22
(Fig. 10). The coil
of the inductance density to 219

4094

IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 11, NOVEMBER 2008

Fig. 10. Comparison of measurement and HFSS simulation data between the
standard solenoid inductor with different scaling factors: (a) inductance and (b)
inductance density.

1
(1 1 )

Fig. 11. (a) Dependences of R on L for different frequencies and (b) comR= L with the calculation data. Dashed
parison of the measured slopes
lines in Fig. 11(a) are the linear fits to the measurement data.

resistance is not affected by the scale-down and is measured to


at 1 MHz. Good agreement is observed
be 0.57 for
between the measurement results and the HFSS simulation data,
with
and the inductance density can reach above 500
if the lateral parameters are reduced by a factor of
three (scaling
) as shown in Fig. 10(b). The tradeoff between the inductance gain and the resistance is again observed
in Fig. 9.
IV. TRADEOFFS IN MAGNETIC INDUCTORS
The tradeoff between the inductance gain and the resistance
increase is quantitatively analyzed in Fig. 11. The inductance
, and the resistance increase,
gain,
, are obtained from various inductor
devices including standard, series and scaling
solenoid
is proportional to
with
inductors. According to (10),
the proportional constant of
. Hence, for a given freand
and
quency, a linear relation is expected between
and
inis confirmed in Fig. 11(a). The slope between
creases faster than the frequency, because the ratio
also
increases significantly with the frequency. The slopes are estimated for various frequencies and compared with the calcu, in Fig. 11(b). The calculation data are
lated values,
obtained using the permeability spectra of the processed magnetic core shown in Fig. 5(a). The measurement results agree
very well with the calculation data, indicating that (10) can accurately describe the tradeoff.
This tradeoff enables us to understand the observed quality
factor of the magnetic inductors. Fig. 12(a) shows the measured

Fig. 12. (a) Measured quality factor data of the standard solenoid inductors and
(b) comparison of the calculated resistance and quality factor with the measurement results for the standard magnetic solenoid with N
:.

= 17 5

quality factors of the standard solenoid inductors. At low frequencies the quality factor is larger for the higher number of
turns because of the larger inductance enhancement, but it starts
to decrease sooner and faster due to the larger resistive losses at
higher frequencies. Nonetheless, the quality factor is above 6 at
turns, and the improvement over the
2030 MHz for
air core case is more than 10 at the peak frequency. An even
higher quality factor can be attained with the use of a thicker
magnetic core, a laminated magnetic core structure or a magnetic material with lower magnetic losses [23]. Resistance and
quality factor for the standard magnetic solenoid with
are calculated using (10) and (12), respectively, and they are
compared with the measurement data as shown in Fig. 12(b).
Again the permeability spectra of the processed magnetic core
shown in Fig. 5(a) are used for the calculations. The excellent
agreements between the calculation and measurement results directly confirm the validity of the proposed analytical model.
Results so far clearly demonstrate that the inductance gain
due to the use of magnetic core comes with the cost of introducing the magnetic power losses at high frequencies. This
increases the device resistance at high frequencies and results
in the drop of the quality factor at high frequencies. For a given
, a larger inductance gain at low
ratio of permeability
frequencies leads to a lower peak frequency for the quality
factor as illustrated in Fig. 12(a). This can be understood as the
tradeoff between the inductance gain and the useful bandwidth
of the magnetic inductor. When the frequency is beyond this
bandwidth, the quality factor of the magnetic inductor becomes
smaller than that of the air core inductor. The analytical models
when the
suggest that this occurs at the frequency
quality factor of the air core inductor is equal to the ratio of
, if the magnetic power loss is the dominant
permeability
loss mechanism at the frequency of interest. Hence,
can be considered as the useful bandwidth of the magnetic
inductor.
V. CONCLUSION
We have designed and fabricated high performance magnetic
inductor devices based on the in-depth understanding of the integrated magnetic inductors and magnetic core materials. For the
,
coil resistance less than 1 and the device area below 1
the inductance as high as 70.4 nH was obtained, corresponding
to the inductance enhancement of 34 over the air core equiv. The
alent, and the inductance density reached 219

LEE et al.: FABRICATION AND ANALYSIS OF HIGH-PERFORMANCE INTEGRATED SOLENOID INDUCTOR WITH MAGNETIC CORE

device properties of the integrated magnetic inductors can be


well understood with the proposed analytical models and hence
can be further optimized for applications and frequency ranges
of interest.
The integrated inductor with magnetic core shows a great
potential for various applications such as the integrated filters,
power conversion, and electromagnetic interference (EMI)
noise reduction. The integrated magnetic inductor can be especially ideal for power delivery applications whose switching
frequency is about 10 MHz or below [24]. However, due to the
intrinsic magnetic losses and the fundamental tradeoffs of the
magnetic inductors, the usefulness of the magnetic inductor
is severely constrained at frequencies beyond 1 GHz. Novel
magnetic materials and inductor designs with much reduced
magnetic losses at GHz frequency range would be needed to
extend the useful bandwidth of the magnetic inductor further
into the GHz frequencies.
ACKNOWLEDGMENT
This work was supported in part by Intel Corporation, National Science Foundation, and Western Institute of Nanoelectronics. The authors would like to thank the staff at the Stanford
Nanofabrication Facility for helpful discussions on the inductor
fabrication and EEJA America for the help with copper electroplating processing.
REFERENCES
[1] S. S. Mohan, M. M. Hershenson, S. P. Boyd, and T. H. Lee, Simple
accurate expressions for planar spiral inductances, IEEE J. Solid-State
Circuits, vol. 34, pp. 14191424, Oct. 1999.
[2] R. F. Soohoo, Magnetic thin film inductors for integrated circuit applications, IEEE Trans. Magn., vol. 15, pp. 18031805, Nov. 1979.
[3] A. M. Crawford, D. S. Gardner, and S. X. Wang, High-frequency microinductors with amorphous magnetic ground planes, IEEE Trans.
Magn., vol. 38, pp. 31683170, Sep. 2002.
[4] L. Li, D. W. Lee, S. X. Wang, K.-P. Hwang, Y. Min, M. Mao, T.
Schneider, and R. Bubber, Tensor nature of permeability and its effects in inductive magnetic devices, IEEE Trans. Magn., vol. 43, pp.
31683170, Jun. 2007.
[5] B. Orlando, R. Hida, R. Cuchet, M. Audoin, B. Viala, D. Pellissier-Tanon, X. Gagnard, and P. Ancey, Low-resistance integrated
toroidal inductor for power management, IEEE Trans. Magn., vol.
42, pp. 33743376, Oct. 2006.
[6] D. S. Gardner, G. Schrom, P. Hazucha, F. Paillet, T. Karnik, S. Borkar,
R. Hallstein, T. Dambrauskas, C. Hill, C. Linde, W. Worwag, R.
Baresel, and S. Muthukumar, Integrated on-chip inductors using
magnetic material, J. Appl. Phys., vol. 103, p. 07E927, Apr. 2008.
[7] C. H. Ahn, Y. J. Kim, and M. G. Allen, A fully integrated planar
toroidal inductor with a micromachined nickel-iron magnetic bar,
IEEE Trans. Comp. Packag. Manuf. Tech., vol. 17, pt. A, pp. 463469,
Sep. 1994.

4095

[8] T. M. Liakopoulos and C. H. Ahn, 3-D microfabricated toroidal planar


inductors with different magnetic core schemes for MEMS and power
electronic applications, IEEE Trans. Magn., vol. 33, pp. 36793681,
Sep. 1999.
[9] D. W. Lee, K.-P. Hwang, and S. X. Wang, Design and fabrication of
integrated solenoid inductors with magnetic cores, in IEEE 58th Elec.
Comp. Tech. Conf. (ECTC), May 2008.
[10] T. H. Lee, The Design of CMOS Radio-Frequency Integrated Circuits,
2nd ed. New York: Cambridge University Press, 2004.
[11] Ansoft Student Licensing Program. Pittsburgh, PA: Ansoft Corp.,
2007.
[12] D. W. Lee and S. X. Wang, Effects of geometries on permeability
spectra of CoTaZr magnetic cores for high frequency applications, J.
Appl. Phys., vol. 103, p. 07E907, Apr. 2008.
[13] D.-X. Chen, E. Pardo, and A. Sanchez, Demagnetizing factors for
rectangular prisms, IEEE Trans. Magn., vol. 41, pp. 20772088, Jun.
2005.
[14] R. C. OHandley, Modern Magnetic Materials: Principles and Applications. New York: Wiley, 1999.
[15] E. V. D. Riet and F. Roozeboom, Ferromagnetic resonance and eddy
currents in high-permeable thin films, J. Appl. Phys., vol. 81, pp.
350354, Jan. 1997.
[16] R. F. Harrington, Time-Harmonic Electromagnetic Fields. New
York: McGraw-Hill, 1961.
[17] M. Yamaguchi, M. Baba, K. Suezawa, T. Moizumi, K. I. Arai,
A. Haga, Y. Shimada, S. Tanabe, and K. Itoh, Improved RF
integrated magnetic thin-film inductors by means of micro slits and
surface planarization techniques, IEEE Trans. Magn., vol. 36, pp.
34953498, Sep. 2000.
[18] M. Yamaguchi, S. Yabukami, and K. I. Arai, A new 1 MHz2 GHz
permeance meter for metallic thin films, IEEE Trans. Magn., vol. 33,
p. 3619, 1997.
[19] E. P. Vandamme, D. M. M. P. Schreurs, and G. van Dinther, Improved
three-step de-embedding method to accurately account for the influence of pad parasitics in silicon on-wafer RF test-structures, IEEE
Trans. Electron Dev., vol. 48, pp. 737742, Apr. 2001.
[20] L. Li, Nanogranular Soft Magnetic Material and on-Package Integrated Inductors, Ph.D dissertation, Stanford University, Stanford,
CA, 2007.
[21] S. C. O. Mathuna, T. ODonnell, N. Wang, and K. Rinne, Magnetics
on silicon: An enabling technology for power supply on chip, IEEE
Trans. Power Electron., vol. 20, pp. 585592, May 2005.
[22] D. W. Lee and S. X. Wang, Multiple magnetic resonances in permeability spectra of thick CoTaZr films, J. Appl. Phys., vol. 99, p.
08F109, Apr. 2006.
[23] L. Li, D. W. Lee, M. Mao, T. Schneider, R. Bubber, K.-P. Hwang, Y.
Min, and S. X. Wang, High-frequency responses of granular CoFeHfO
and amorphous CoZrTa magnetic materials, J. Appl. Phys., vol. 101,
p. 123912, Jun. 2007.
[24] A. Ghahary, Fully integrated DC-DC converters, Power Electron.
Technol., pp. 2427, Aug. 2004.

Manuscript received March 03, 2008; revised July 30, 2008. Current version published December 17, 2008. Corresponding author: D. W. Lee (e-mail:
dwlee@stanford.edu).

Вам также может понравиться