Gross R., Sidorenko A., Tagirov L. Nanoscale Devices

Подождите немного. Документ загружается.

114 G. Panaitov, R. Ott, N. Klein

tan 0

ββπ

=⇒ = ⇒=lll.

(5)

In this case, the closed end slotline operates as a half-wave resonator.

Using a MEMS switch at the right end of the slotline one can control the

load of the slotline and, as the consequence, to change the resonant

frequency of the slotline. We show below that this behavior can be used for

development of a frequency tuning mechanism for the dielectric resonator.

MEMS

Microelectromechanical switches are known to demonstrate very useful

performance at microwave frequencies [1]. Advantages of the MEMS over

the conventional semiconductor switches are low losses, low consumption

and a higher “off/on” resistance ratio. Figure 2 shows an example of

MEMS switches, which we develop for application in the tuneable

resonator. A sacrificial layer technique used to fabricate a free-standing Au

microbridge structure. The two microstrip lines underneath supposed to be

contacted by bending of the microbridge structure due to actuation of the

MEMS. A Nb layer on the top of the Au microbridge illustrates a bimetal

thermo-expansion actuation mechanism.

Fig. 2. Microelectromechanical switch design.

Tuning approach

Our concept of a MEMS tunable dielectric resonator is illustrated in Fig. 3.

A substrate with four radial slotlines (white strips, Fig. 3) is symmetrically

arranged

above the cylindrical dielectric resonator of TE

01δ

mode. The

inner end of each slotline is a short, while outer end is loaded by a MEMS

switch (black strips). For the case of a MEMS off state, each of 4 slots

represent a quarter wave slotline resonator, which possess a strong

magnetic field component in radial direction providing strong coupling to

MEMS Tunable Dielectric Resonator 115

the resonant fields of the TE

01δ

mode. Upon closing a single MEMS, the

corresponding quarter wave resonator changes into a half-wave resonator.

Due to the sign change of magnetic field in the middle of the slotline, the

coupling between slotline and TE

01δ

mode is much weaker. The variation of

the coupling between two resonance modes leads to change of the

resonance frequency of the assembly.

Fig. 3. Schematic of the MEMS tunable resonator. The substrate with 4 slotlines

(white strips) is symmetrically arranged above the cylindrical dielectric resonator at

distance h. Black strips indicate MEMS position at the end of slotlines.

Due to resonant interaction between two resonance modes the stored

electromagnetic energy in the MEMS open state is distributed between the

DR and slotlines. This leads to an increase of the effective size of the

resonator and, as a consequence, to decrease of resonant frequency of the

01δ

mode. In the MEMS closed state the intermodal coupling is much

weaker, the electromagnetic energy is concentrated within a DR and

resonance frequency is about of non-disturbed DR.

The tuning range of the resonator is determined by the coupling between

the TE

01δ

mode and the quarter wave slotline mode, which can be controlled

by a variation of the distance h (Fig. 3). The major advantage of the

technique is that in the MEMS closed mode there is almost no coupling

between the dielectric resonator and the slotline, i.e. one can expect only

small additional losses due to the MEMS structure. In particular, the

contribution of contact resistance of the closed MEMS should be rather

small.

116 G. Panaitov, R. Ott, N. Klein

Numerical Field Simulations

In order to evaluate the expected tuning range and to choose an optimum

slotline configuration, numerical field simulations have been performed

employing the commercial code “CST Microwave Studio”.

For the resonator design shown in Fig. 3, the spatial distribution of the

resonant fields was calculated for the capacitive (open) and resistive

(closed) state of the MEMS. Fig. 4 illustrates the contours of rf magnetic

field in the meridian plane for both two states. The black contours

correspond to the higher field amplitude. As expected, the field distribution

differs substantially. There is an obvious excitation of the slotlines with

open end (Fig. 4a). For the capacitive state the resonance frequency was

found to be at f = 1.878 GHz, the quality factor is limited to about of Q =

6.700 due to losses in the magnetic field sections of the slotlines.

Fig. 4. Magnetic field contours in the meridian plane at: a) quarter-wave mode,

MEMS in the capacitive state; b) half-wave mode, MEMS in the resistive state.

For the resistive state, the field is localized primarily inside the

dielectric resonator (Fig. 4b), therefore the effects of slotlines and MEMS

are not significant. Nevertheless, the quality factor of Q = 21.500 is still

slightly lower than that of the non-perturbed dielectric resonator. This is

due to a weak excitation of the half-wave slotline mode and corresponding

losses in the resistive state. The excitation of the half-wave mode is

obviously seen in the Fig. 5b. Here is the distribution of the electric field in

the

meridian plane of the resonator at half-wave mode. The two

maximums within the DR contour indicate that the electric field has an

azimuthal symmetry. One notices that the amplitude of the quarterwave

field (Fig. 5a) in the slotline area is about one order of magnitude larger

than

that of the half-wave mode. Figures 4-5 show that both magnetic and

a) b)

MEMS Tunable Dielectric Resonator 117

electric fields are substantially different in two resonator modes.

Therefore, the variation of the resonance frequency by switching of MEMS

is a result of a complex variation of both fields, according to (1). For the

parameters taken in the calculation the tuning range is about 61 MHz, i.e.

3.2%.

Fig. 5. Electric field contours in the meridian plane at: a) quarter-wave mode; b) at

half-wave mode.

Experimental Results

The principle of tuning by MEMS has been proven in an experimental

setup, which is based on a commercial piezoelectric bimorph actuator. The

details of the experimental design are shown in Fig. 6. A cylindrical

dielectric resonator (

= 28, BZT ceramic, by TransTech) with TE

01δ

mode

at about f = 1.86 GHz is arranged in a cooper cavity. A substrate with four

radial slots was assembled coaxially above the dielectric resonator. As

MEMS equivalents, four metallized bimorph piezoelectric actuators control

the slotlines states and consequently, the coupling between the DR

a) b)

118 G. Panaitov, R. Ott, N. Klein

Substrate with slotlines

Bimorph actuator

HF Coupling

ctuation Supply

Fig. 6. Design of the tunable resonator with piezo actuators as the MEMS

equivalents.

and the slotlines. The maximum deflection of actuators is about ±330

with a quite high blocking force of 2.5 N at its tip. The latter is very

important in order to achieve a high contact resistance for the slotline

closed state. A compact bipolar ±100 V power supply has been developed

in order to actuate the bimorph. In addition, the bimorphs respond quite fast

(~1.5 ms switching time) and they are very stable with specified cycling up

to 10

. The properties of single slotlines have been tested separately in a

rectangular cooper cavity. The resonant frequency of the slotlines was

found to be correlated with the slots length according to (4). The quality

factor of the slotline, depending on the width of the slots, was evaluated to

be in the range of 200 ÷ 400 at f ~ 2 GHz. Four slotline disks based on

metal thin films deposited on sapphire wafers and bulk copper disks with

different slot size have been tested in resonator design (Fig. 4) to achieve an

optimum tuning. Table 1 shows the frequency and quality factor (Q) of the

resonator depending on the slotline state for four bulk cooper slotlines at

two distances above the DR.

MEMS Tunable Dielectric Resonator 119

Table 1. Parameters of the resonator with slotlines of 19×0.4×0.5 mm

size at two

distances, h, between slots and dielectric resonator.

h = 2.5mm h = 12mm

Slots state

f, MHz Q f, MHz Q

4 closed 1917.2 11500 1856.26 25800

3 closed 1913.3 8500 1856.00 25000

2 closed 1909.1 8000 1855.72 25000

1 closed 1906.3 7000 1855.45 26800

0 closed 1902.9 6000 1855.20 26500

For slotlines of 19×0.4×0.5 mm

size the highest Q was achieved in the

case of four slotlines being closed. Upon increasing the number of open

slotlines, Q decreases gradually down to a minimum for all slotlines being

in the open state. The most interesting result is that the resonant frequency

changes step-kind by a discrete almost constant value. The mean frequency

step is about 3.5 MHz and 0.25 MHz per actuator switch for two distances.

The discrete variation of the frequency by switching of the slotline state

means that there is no substantial intermodal interaction between slotlines.

This behavior can be used for a digital frequency tuning. Any required

tuning range

= n

f can be realized by actuation of an appropriate

number, n, of tuning units (slots) with a frequency step

f. In order to

minimize the number of tuning elements while keeping the required tuning

range one can even take slotlines with two different frequency steps

and

. For example, in order to realize a tuning within a 5 MHz range with the

frequency step

f = 0.2 MHz one needs just 8 slotlines with frequency steps

= 0.2 MHz and

= 1 MHz, instead of 25 slotlines in case of a single

frequency step of

f = 0.2 MHz.

In order to further increase the Q,-factor of the resonator we have tested

broader slotlines (w = 1 mm), which are known to have a higher Q [4].

However, the Q-values for the closed wide slots have been not as high as

expected. One possible reason for this might be the additional dielectric

losses in the actuator material. These losses may be reduced by a

metallization of the entire surface of the actuator. As the main difference,

the achieved frequency step per slotline is 7 MHz, which is twice as high as

for the 0.4 mm slotlines. Another possibility to change the tuning range is

variation of the distance between the DR and slotlines.

120 G. Panaitov, R. Ott, N. Klein

24681012

f, 19x1x0.5mm

f, 19x0.4x0.5mm

f, 12x1x0.5mm

Frequency/switch, MHz

h-distance, mm

Fig. 7. Frequency step, δf, as the function of the h-distance between DR and slots

of three different sizes: 19×1×0.5 mm

19×0.4×0.5 mm

and 12×1×0.5 mm

function of separation between the slotline disk and the dielectric resonator,

h, for different slotline size. Thus, depending on the application one can

choose an appropriate slot size to achieve a specified frequency tuning step.

It is important to note that the figure-of-merit for tuning,

⋅ ,

is not constant for the different slotline size, and therefore may further be

optimized. An example of measurements of resonance parameters as the

function of the slotline length is shown in Fig. 8. The measured data

illustrate that there is a strong compromise between tuning range and

quality factor. The quality factor drops down with the length of the slotline

due to additional losses in the slots. Further investigations are directed

towards replacing the slotline resonators by other MEMS-tunable planar

resonator with higher Q. In any case, a switching of the resonance

frequency by about 40 MHz is observed with quality factors being about

7000. This result corresponds to the figure-of-merit values of about 160,

which is very challenging for tunable resonators.

Figure 7 shows a series of experimental dependences of the frequency as the

MEMS Tunable Dielectric Resonator 121

10 12 14 16 18 20

Frequency/switch, MHz

Slotline length, mm

f, MHz

8000

9000

10000

11000

12000

Quality Factor

Conclusion

We proposed a concept of fast tuning of a bulky dielectric resonator

coupled to planar slotline resonator. It was shown that MEMS tuning of

dielectric resonators is quite challenging and has a great potential to beat

the performance of semiconductor based tuning techniques. The discrete

variation of the frequency by switching of the slotline state can be used for

a digital frequency tuning. Optimization of the planar resonator structures,

in order to reach the larger tunability while keeping the highest quality

factor, is a subject of our future activity.

References

1. Rebeiz GM, Muldavin JB (2001). Revue HF (Belgium) 2:39-52

2. Klein N (2003) “Microwave Communication Systems: Novel Approaches for

Passive Devices”. In: Waser R (ed) Nanoelectronics and Information

Technology, Wiley, New York

3. Kajfez D (1986). In: Kajfez D, Guillon P (eds) Dielectric resonators. Artech

House, Dedham, MA

4. Rozzi M et al. (1990). IEEE Trans MTT-38:1069-1078

Fig. 8. Frequency step and quality factor as the function of slotline length.

Ultra-Thin Spin-Valve Structures Grown on the

Surface-Reconstructed GaAs Substrate

B. Aktaş

, F. Yıldız

, O. Yalçın

, A. Zerentürk

, M. Özdemir

L.R. Tagirov

, B. Heinrich

, G. Woltersdorf

, R. Urban

Gebze Institute of Technology, 41400 Çayırova-Gebze, Turkey

Kazan State University, Kazan 420008, Russia

Simon Fraser University, Burnaby, BC, V5A 1S6, Canada

Abstract: We studied magnetic anisotropies of epitaxial, ultra-thin iron spin-

valve structures grown on surface reconstructed GaAs substrates. The

ferromagnetic resonance (FMR) technique has been exploited to

determine magnetic parameters of the ferromagnetic films in the

temperature range 4-300 K. The unusual and unexpected angular

dependence of FMR spectra allowed us to build and verify a model

describing the magnetic anisotropy of the studied systems. Switching

of the principal anisotropy axes of the second layer has been

observed. It was attributed to a drastic relaxation of the uniaxial

component of anisotropy induced by the surface reconstruction of the

substrate. The linear variation of magnetic anisotropy parameters

with temperature has been observed. The results on temperature

dependence are discussed in terms of thermal expansion induced

magneto-elastic anisotropies.

Keywords: ferromagnetic resonance, magnetic anisotropy, ultra-thin film spin-

valve

Introduction

The magnetic anisotropy of thin films is of crucial importance in spintronic

applications. It is well known that ferromagnetic resonance (FMR) is the

most sensitive and accurate technique to determine magnetic anisotropy

fields of very thin magnetic films [1, 2]. In this paper we study the

123

R. Gross et al. (eds.), Nanoscale Devices - Fundamentals and Applications, 123–134.

124 B. Aktaş, F. Yıldız, O. Yalçın, et al.

magnetic anisotropies in iron bi-layer structures grown on the surface-

reconstructed GaAs single-crystalline substrates and demonstrate how

surface-induced anisotropy can be used to tailor overall magnetic properties

of the studied spin-valve structure. In our experiments we observed

unconventional and unexpected FMR spectra. They are interpreted and

explained based on the model proposed in this study. Switching of the

principal anisotropy axes of the second layer with respect to the first layer

has been observed. Consistent fitting of angular and frequency

dependencies of the FMR spectra in the temperature range 4-300 K allowed

us to determine accurately the cubic, uniaxial and perpendicular

components of the magnetic anisotropy, as well as establish directions of

easy and hard axes for the magnetization in the layer(s). The origin and

temperature dependence of the magnetic anisotropy fields are extensively

discussed in terms of the surface-induced anisotropy and lattice mismatch

between materials subject to stress induced by differences in thermal

expansion coefficients of materials in a contact.

Samples and general FMR measurement procedure

Single and double layers of ultra-thin iron films were prepared by

Molecular Beam Epitaxy (MBE) on (4×6) reconstructed GaAs(001)

substrates. A brief description of the sample preparation procedure is as

follows. The GaAs(001) single-crystalline wafers were sputtered under

grazing incidence using 600 eV argon-ion gun to remove native oxides and

carbon contaminations. Substrates were rotated around their normal during

sputtering. After sputtering the GaAs substrates were annealed at

approximately 500°C and monitored by means of reflection high energy

electron diffraction (RHEED) until a well-ordered (4×6) reconstruction

appeared [3]. The (4×6) reconstruction consists of (1×6) and (4×2) domains

with the (1×6) domain being As-rich and the (4×2) domain Ga-rich.

The

Fe films were further deposited directly on the GaAs(001)

substrates at room-temperature from a resistively heated piece of Fe at the

base

pressure of 1×10

-10

Torr. The film thickness was monitored by a

quartz

crystal microbalance and by means of RHEED intensity

oscillations. The deposition rate was adjusted at about one mono-layer

Experimental Result

Gross R., Sidorenko A., Tagirov L. Nanoscale Devices - Fundamentals and Applications

Подождите немного. Документ загружается.