         

 

 

1 LIBRARY
”700 Michigan State
0“ W University

 

 

 

This is to certify that the
thesis entitled

A WIRELESSLY CONTROLLED MAGNETIC FORCE
ACTUATOR AND ITS AFM MEASUREMENTS

presented by

Faisal T. Abu-Nimeh

has been accepted towards fulﬁllment
of the requirements for the

MS. degree in Electrical and Computer

 

 

Engineering
é WAR/Q

\' ‘7 Major Professor’s Signature

Max; 15, 906‘7

Date

MSU is an afﬁrmative-action, equal-opportunity employer

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:lCIRC/DateDue.indd-p.1

 

A WIRELESSLY CONTROLLED MAGNETIC FORCE
ACTUATOR AND ITS AFM MEASUREMENTS

By

Faisal T. Abu-Nimeh

A THESIS

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Electrical and Computer Engineering

2007

ABSTRACT

A WIRELESSLY CONTROLLED MAGNETIC FORCE
ACTUATOR AND ITS AFM MEASUREMENTS

By

Faisal T. Abu-Nimeh

Nanotechnology enables imaging and manipulating objects at the nanoscale possible.
However, the majority of instruments and tools used in Nanotechnology include bulky
equipment and require large voltages. In particular, actuation is typically achieved
via Piezo material that requires voltages over the hundred volt range. The present
thesis describes a portable low—power (low-voltage) magnetic force actuator mod-
ule controlled wirelessly using the Zigbee protocol and its Atomic Force Microscopy
(AFM) measurements. The contribution of the study is three fold: ﬁrst, we design,
simulate, and test a chip, which produces eight different magnetic force levels us-
ing the AMI uO.5m CMOS technology. All circuit components are implemented on
the same die by minimizing their sizes and also minimizing the number of I/ O pins.
Thus, the circuit design real-estate size is 535nm x 378um. Second, we implement
an embedded software to wirelessly control the actuator chip via the Zigbee proto—
col. The software implements two command methods; either an automated command
by executing a preprogrammed algorithm stored on a microcontroller, or a manual
command by processing the user’s input during runtime. Third, we measure the
nano—Newton electromagnetic (EM) forces generated using the AFM by placing a
magnetically doped cantilever (MFM: Magnetic Force Mode) over of the actuator

platform and monitor its deﬂection for different force generated (EM) levels.

To my parents.

iii

ACKNOWLEDGMENTS

First, I would like to thank my family for their great support and guidance.
Despite the great distance between us they have always been there for me throughout
my educational journey.

I would like to thank Professor F athi Salem for his kind efforts and help since the
beginning of my graduate studies. I have learned a lot during my work in his lab.
His vast experience and enlightening advice made this thesis work possible. Also,
many thanks to Professor Shantanu Chakrabartty and Professor Subir Biswas for
their assistance and help throughout my studies.

Finally, thanks to all faculty, staff, and students at Michigan State University
speciﬁcally those in the Electrical Engineering Department for making this a great

institution and learning environment.

iv

TABLE OF CONTENTS

LIST OF TABLES .............................. vii
LIST OF FIGURES ............................. viii
1 Introduction ................................ 1
1.1 Background and Motivation ....................... 1
1.2 Design Speciﬁcations ........................... 2
1.3 Thesis Organization ............................ 3

2 CMOS Magnetic Force Actuator ................... 5
2.1 Introduction ................................ 5
2.2 Digitally Controlled Class E Power Ampliﬁer .............. 5
2.2.1 Introduction ............................ 5

2.2.2 Class E Power Ampliﬁer ..................... 6

2.2.3 Digitally Controlled Switches .................. 12

2.2.4 Thermometer Decoder ...................... 14

2.2.5 Magnetic Force Actuator ..................... 17

2.3 Conclusion ................................. 21

3 Wireless Control ............................. 23
3.1 Introduction ................................ 23
3.2 Zigbee Protocol .............................. 24
3.2.1 Introduction ............................ 24

3.2.2 Zigbee devices ........................... 24

3.2.3 Zigbee Stack ............................ 27

3.3 Wireless Nodes .............................. 28
3.3.1 Controller (Coordinator) ..................... 28

3.3.2 End Devices (MFA Nodes) .................... 31

3.4 Conclusion ................................. 33

4 Atomic Force Microscope ........................ 34
4.1 Introduction ................................ 34
4.2 Magnetic Force Measurements ...................... 36
4.2.1 Introduction ............................ 36

4.2.2 Measurements ........................... 36

4.2.3 Conclusion ............................. 4O

5 Conclusions ................................ 43

5.1 Results ................................... 43
5.2 Future Work ................................ 46
APPENDICES ................................ 47
A Magnetic Force Actuator Test Chip .................. 47
BIBLIOGRAPHY .............................. 50

vi

2.1
2.2

3.1
3.2

4.1
4.2
4.3

LIST OF TABLES

Simulation Results for Output Power with Different Inputs ....... 14
Thermometer Decoder INPUT/ OUTPUT ............... 15
Zigbee Operational Frequency, Bit Rate, and Number Of Channels . 24
Manual User Input Using Switches ................... 30
Comparison Between AFM and Other Technologies .......... 36
Different User Inputs with their Corresponding Deﬂections ...... 39
Different User Inputs with their Corresponding Forces ......... 40

vii

2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
2.14
2.15
2.16
2.17
2.18
2.19

3.1
3.2
3.3
3.4
3.5
36

LIST OF FIGURES

Class E Power Ampliﬁer ......................... 6
Optimum waveforms for maximum efﬁciency[1] ............. 8
60nH Spiral Inductor ........................... 9
Class E PA when M1 is ON ....................... 9
30nH Spiral Inductor ........................... 10
Class E PA when M1 is OFF ...................... 10
1pF poly-poly2 Capacitor ........................ 12
8-bit Digitally Controlled Class E PA .................. 13
Output Power 8-bit Digitally Controlled Class E PA .......... 14
VHDL Code for Thermometer Decoder 3-to-8 ............. 16
3~to—8 Thermometer Decoder Schematic ................. 16
A 6 micron by 6 micron Actuator .................... 17
Unstacked Metal Layers for The Actuator ............... 17
3D Angle View for The Magnetic Force Actuator ........... 18
3D Side View for The Magnetic Force Actuator ............ 18
AMI 0.5 Metals and VIA Properties ................... 19
Current Passing Through The 20pH Inductor ............. 20
Final Schematic for The Magnetic Force Actuator ........... 21
Final Layout for The Magnetic Force Actuator ............. 22
Magnetic Force Actuator in A Star Network Conﬁguration ...... 25
Zigbee Protocol Stack Architecture [2] ................. 27
Coordinator Interrupt Routine to Start Actuation ........... 29
Coordinator User Input Manual Actuation ............... 30
Coordinator Preprogrammed Auto Actuation ............. 31
RFD Node Receiving 3—Bits That Represent 8 Levels ......... 32

viii

4.1 Atomic Force Microsc0pe [3] ....................... 35
4.2 Tip Deﬂection at 111 ........................... 38
4.3 Tip Deﬂection at 110 ........................... 39
4.4 Tip Deﬂection at All Input Levels .................... 40
4.5 Force Exerted On Tip at All Input Levels At Distance Z1 ....... 41
4.6 Magnetic Cantilever on Top Of MFA .................. 42
5.1 Magnetic Force Actuator Chip ...................... 43
5.2 Force Exerted On Tip at All Input Levels At Distance Z1 ....... 44
5.3 Magnetic Cantilever on Top Of MFA .................. 44
5.4 Thesis Block Diagram .......................... 45
A1 MFA Test Chip Layout .......................... 47
A2 MFA Fabricated Test Chip ........................ 48
A3 MFA Pin Diagram ............................ 49

ix

CHAPTER 1

Introduction

This chapter provides a brief introduction of the overall thesis, starting with the
background and motivation, then going over the design challenges and speciﬁcations,

and ﬁnally ending up with the organization of this thesis.

1.1 Background and Motivation

As engineers, we are instructed and taught how to design things, however, any design
by itself has different levels. Starting at the bottom: the device level is considered the
lowest [4] then the circuit, gate, module, system, and all the way up to the application
level. This work involves many levels of design. It starts with the device level to build
up a system that is fabricated on a chip, which is later used with other components

to deliver a ﬁnal design. The three major components of this design are:
0 A fabricated chip that contains a magnetic force actuator.
o Programmed wireless nodes to communicate with the chip mentioned above.
0 An Atomic force microscope (AF M) to sense and measure the actuated force.

The thesis work integrates these components into a working system and executes

experiments to demonstrate the system’s functionality.

The on-chip actuator generates eight different levels of electromagnetic forces in
the nano ranges suitable for nanotechnology applications and it uses low voltage power
supplies (5 volts for the AMI 0.5um), in contrast with Piezo material based nano-
poistioning actuators which require very high voltages (100+ volts) [5]. Therefore,
the proposed low voltage actuator can be portable and thus wireless command and
control becomes an option. On the other hand, several applications require precise
positioning, thus demand nano accuracy for their forces. For example, in lithography
and semi-conductor fabrication processing, precise incremental forces are required. In
[6] a motor-piezo technique is used for nano scale precision but that requires extra
elements to compensate for nonlinearities and hysteresis. Whereas in this work, an
integrated 8-bit digital input provides incremental and stable eight different levels of
actuation to provide the desired forced. Consequently, the size of the actuator, the
way it is controlled, the low power requirements, and incremental digital actuation
make the proposed work more attractive for nano applications.

Measuring and validating the nano-Newton level forces becomes a challenge for
present day measuring devices. In this work, an AFM is used to quantify the CMOS

silicon generated force for accuracy and resolution.

1.2 Design Speciﬁcations

Each component mentioned in the previous section has its own speciﬁcation. Starting

with the fabricated chip, its design requirements are:
0 The size of the chip (area) has to be as small as possible.
0 An easy interface to control the chip with near linear behavior.
0 All components have to be integrated on one die.

0 The power consumed by the chip has to be as minimum as possible.

The second part of this work deals with programming wireless nodes to communicate

with the chip and it is required to have the following speciﬁcations:
0 Two or more wireless nodes are for communication and / or control.
a The size of the code should be minimal.
0 The algorithm used to control the chip is computed and executed remotely.
o The wireless nodes should not consume much power.

The last part, which is the AFM, is required to have the following speciﬁcations:
0 The ability to measure small magnetic forces.

0 The cantilever tip should be small enough to access the small elements on the

fabricated chip.

These were the overall speciﬁcations for this work. An in-depth assessment will be
discussed for each requirement in its corresponding chapter to explain why and how

such requirements were attained.

1 .3 Thesis Organization

The ﬁrst chapter includes a brief introduction of the project. It contains three sec-
tions, the ﬁrst one is background and motivation, the second one lists the design
speciﬁcations, and the third one describes the organization of this thesis.

Chapter two included all material regarding the fabricated chip with detailed
speciﬁcations, simulations, and experimental results.

Chapter three describes how the fabricated chip can be controlled wirelessly using
the Zigbee protocol with examples and real implementations.

Chapter four introduces the AFM module and how it is utilized it to measure the

output of the fabricated chip.

Finally, the last chapter gives an overview of the entire system with all components

and possible future work.

CHAPTER 2

CMOS Magnetic Force Actuator

2.1 Introduction

This chapter discusses the design process and results for the fabricated CMOS chip
that contains a Class E power ampliﬁer, a Thermometer Decoder, and a Magnetic
Field Actuator. Each component will be explained separately and in the last section
a summary of the whole chip will be given. The current CMOS technology used for
this project is the AMI C5N n—well 0.5a and 0.30A technology [7] which features 3
metal layers, 2 poly layers, and a high resistance layer. The supply voltage used with
this technology is 5 volts, consequently all circuits will be designed and optimized for

5V.

2.2 Digitally Controlled Class E Power Ampliﬁer

2.2.1 Introduction

Power Ampliﬁers (PA’s) are one of most important elements in a CMOS transceiver.
Its power consumption is signiﬁcant to the overall total power used. Thus, making
it as efficient as possible yields great savings specially for low power applications.
Achieving high efﬁciency in CMOS is limited by low breakdown voltage, low cur-

rent drives, substrate noise, etc. However, there are techniques to go around them.

Another important requirement is linearity, which makes the usage, design, and oper-
ation easier. The PA included in this work is similar to that designed in [8], however,

with few minor changes it can be exploited for different applications.

2.2.2 Class E Power Ampliﬁer

The ﬁrst generation of the nonlinear Class E PA appeared in the mid 70’s when
Soka1[1] introduced a highly efficient PA with a 100% ideal efﬁciency. The Class E
ampliﬁer relies on the fact that the active device, FET or BJT, operates as a switch
instead of current source. It became very attractive due to its simple design and
robust behavior. Figure 2.1 is a simple Class E PA and it contains three major
elements: the transistor M1 which acts as a switch, a ﬁlter (C2,L2) which tunes the
output to a certain frequency, and a RFC (Radio Frequency Choke) large inductor
that assures that the peak to peak swing at the output load is close to 3.6xVDD and

acts as a DC current source[9].

+5V

RFC

C2 L2 Out

In X _]_l m R
_|._ L

 

 

0

 

.___. C1

N” _I_

 

 

Figure 2.1. Class E Power Ampliﬁer

Ml operating as an ideal switch, meaning that there is no overlapping period of
nonzero switch voltage and current. So, when the switch is ON, the drain voltage at

X is zero and when the switch is OFF, the current ﬂowing into MI is zero. Figure

2.2 shows the ideal wave form for the switch M1 where the current and the voltage
do not intersect at the same time. However, the transition between the ON and OFF
state has a ﬁnite time, so, that’s when losing power occurs. To eliminate this an
abrupt transition between the ON and OFF state is desired. However, in reality this
does not exist, consequently a step function input is used with very small rise/fall
time. Also, for none ideal conditions a shunt capacitor C1 is used to reduce the
switching losses that occur during the transition from ON to OFF to make sure that
the voltage at X increases from zero just after the current stops ﬂowing into M1. On
the other hand, the inductor L2 and the capacitor C2 can also minimize the losses
during the transition from OFF to ON by assuring that voltage goes to zero just
before the switch current increases. The criterion for selecting the shtmt capacitor
C1 is: neither too large to be considered a harmonic short nor too small to become a
device parasitic[10].

The methodology used to design such a PA is straight forward. Starting with the
RFC: inductor, it should provide a DC path from the supply and is approximately an
open circuit at RF[11]. Hence, it has to be large enough to provide enough current,
consequently, a 60nH on-chip spiral inductor was designed and it is shown in ﬁgure
2.3. The dimensions of the spiral are 103nm by 103nm, the width of the metal layers
is 1.8nm, the spacing between metals is 0.9um, the number of turns (windings) is 18,
and ﬁnally the metal layers used are Metal3 and Meta12 (exit). Note that the size,
spacing, and width is as minimal as possible to make sure that we can implement the
smallest functional chip possible using the AMI 0.5u technology. Even though we have
a 1.5mm by 1.5mm real-estate but keeping the overall used area minimal is crucial for
certain future applications. This inductor was designed using ASITIC[12], using the
command in (2.1). The resulting spiral inductor had an inductance approximately

equals to 60nH and a Q value of 1.2.

°VOLYAGE ACROSS swncu

 

A

I
I
I
I
I
I
I
I
I
l
I
I
I
I
l
I
I
I
I
I
I
I
I

 

 

° cunnsur mucosa swrrcn

 

 

inn:

3
]___ swncn . .. swncu' ]
“orﬂsms _ on” STA‘IE

Figure 2.2. Optimum waveforms for maximum efﬁciency[1]

sqmmname = ind60 : len = 103 : w = 1.8 : s = 0.9 : n = 18 : metal = m3: exit: m2
(2.1)
The next part is to tune the PA to run at the desired frequency 900MHz. To
achieve that high frequency two on chip passive elements were designed; a 45.5 pF
capacitor and a 30nH inductor. Therefore, two models characterize this Class E PA:
when the switch is turned ON and the other one is when the switch is turned OFF.
First, when the switch is ON, ﬁgure 2.4 shows the model consisting of a simple
RLC. This part can be clearly used as a band pass to block other frequencies and
to minimize the harmonic distortion caused by other e1ements[13]. This model will

produce a resonance frequency based on equation (2.2) were we is the operational

I ’ f. I.” f/ 7:“) /
x’x'fx'III/lx 'I/x’x'lx'. -'///I//x’/ ’/////x’ // - "/x’xﬁ’xix’f/I/fx‘x‘x’x’x'xV/x'x
V ’1 I I 'x'l/x/{I/I/Il ‘I/I/I/I'

'x’x’l/l/ﬂ’x’x (VI/L"; .
”III

2'
2
3
ﬂ
%
I
3
x,
g
5
I
/,
7
,
g
I
I
/
§
4
i
’x’
f
I
I
ﬁ
I,
ﬁx
5
I
5
f
at
:51
I
/
,4
ﬂ
,5.

. .-»: xxx/xxx
:. ' ‘ ”xx 77/
' 1’1
' . . xxx/xx x/xx
, w; ' ' »4 xxx;
. xxx/xx _.-../xxxx
' j ' ,....x..-'.x xx
'1 - .," ' xxx/x tx .. I
. _ . . ' . xx ’
' ' ‘ - - .' ”xx/”xxx / x
-‘ , h xxx-r / x ' '
'r, x...x/xx .. xx xx
" ,. -: . . xx.
,1. 3 " :_’
' f. . . I
. , . g xxx/xx /
' ,. y ’ f ; .- . xxxx
_: .. 5
. ,
/ .. /
..1 I. '- . ‘
;.-5 _. .' - ' '2 f x/xx/xxx
. , ‘ 2, ' xx.
‘ f; / .- xxx/x
- _ - l // r
; . /
. : ' - xxx
. " . ‘ xxx .'
- . , " .. .xxa’xx ;«x xxx ”xx/”xxx.”
' . x ' x‘x'fllx'l'II/III xxx}! xx xxx/xxx” /
“ I'xx/ '- / "xxxx/ 7/” [x
y t‘ xx ' .xxx xxxx Uxxxx/ xx. xxx/l
:' xx. xxxx xxx/xxx/xx/xxx N I
xxx '- ' -‘ . I “xxx /
xxx/x x/xx

 

/.
r;
9
.4
I
x
.4
5
I
§
(5
f;
;.
f
I
x
x
S"
i
/
.4
x
,4
x
52':
6
x
?
/
x
5
.5
' x/

Figure 2.3. 60nH Spiral Inductor

M1 2 L2 I=IL

Figure 2.4. Class E PA when M1 is ON

frequency.

(do = W (2.2)

To stay within constrains of the technology and real-estate limitations we chose L2
to be 30nH. Again, ASITIC was used to simulate and design this spiral inductor as
shown in ﬁgure 2.5. The dimensions of the spiral are 103um by 103nm, the width of
the metal layers is 1.8um, the spacing between metals is 1.8um, the number of turns
(windings) is 10, and ﬁnally the metal layers used are Metal3 and MetalZ (exit). The

only difference between the 30nH inductor and the 60nH is the spacing used between

9

the metals and the number of turns. By plugging in L2 = 30nH and we = 900MHz
in equation (2.2) then 02 will be 45.5pF.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2.5. 30nH Spiral Inductor

Alternatively, when the switch is OFF the circuit can be represented as shown in

M‘Il

ﬁgure 2.6.

Figure 2.6. Class E PA when M1 is OFF

Using the same values for L2 and 02 from the previous model we ﬁnd C1 to be 41.5pF

using (2.3).
1

= 7L2(02-01)/(C2+01)

 

£01 (2.3)

10

However, this value will reduce the efﬁciency of the circuit. Because when the switch
is OFF the big inductor RFC in ﬁgure 2.1 will start ﬁlling up the Capacitor Cl. So,
by the time the switch is back on, 01 has to get rid of all the charge it has stored
and if it doesn’t then the charge will be wasted by the switch and that will reduce
the efﬁciency of the PA. Therefore, the optimal Cl should be calculated[14] using

equation (2.4).

 

P
01 = a“; (2.4)
where Pout is the power at the output, w is the operational frequency, and VD D is the
power supply voltage. Assuming that the PA is connected to a 500 load, the power

at the output can be estimated with equation (2.5).

2
R (r2 + 4)
By substituting Pout in equation (2.4) with (2.5), 01 will be approximately 1pF. Both

Pout 2 (2'5)

capacitors 01 and 02 has been designed using two consecutive conducting plates of
polysilicon. Using the values from the previous MOSIS runs for AMI 0.5” the capac-
itance between the two layers poly and poly-2 equals c(poly-poly2) = 0.914fF/um2.

To ﬁnd the size of the capacitor we need to calculate the ratio (S) between the desired

DesiredCapacitance
S heeWapacitance

 

capacitance and the sheet unit capacitance. Therefore, 8 = so,
S = [[5154 F = 45.405k, so, the length and the width (because it is a square) is 213nm
(to the nearest A). The same calculation is carried out for the 1pF capacitor and it
yields a length which equals 33.15pm, shown in ﬁgure 2.7.

All the parameters needed for the PA are ready except for the switching part
which will be discussed in the next subsection. To summarize the values, The RFC
inductor is 60nH, shunt capacitor 01 is 1pF, to tune the output frequency a bandpass

ﬁlter with Cg equals to 45.5pF and L2 30nH, and ﬁnally the load R L is considered
to be a 509 load.

11

 

Figure 2.7. 1pF poly—poly2 Capacitor

2.2.3 Digitally Controlled Switches

Horn the previous subsection as shown in ﬁgure 2.1 the switch M1 is a simple NMOS
transistor. Therefore, when we consider the technology we are using, it is important
to operate the switch as fast as possible to accommodate the 900MHz input frequency.
So, to match this requirement, the size of the transistor Ml has to be as small as
possible. Additionally, to increase the power of the PA [15] the transistor Ml can
be replaced with a parallel conﬁguration of NMOS transistors (M1,...,M8). This
conﬁguration also uses the minimum size allowed W/ L ratio to cope with the 900MHz
input frequency.

The eight transistors added to the PA could allow more current to be delivered
to the output, however, adding more transistors to the parallel conﬁguration will not
result in a signiﬁcant change in the current delivered, because the circuit actually
saturates and it can no longer deliver more current to the output.

Having eight transistors as inputs allows us to have eight different power output

12

levels by switching transistors ON / OFF when needed. Therefore, an AND gate is
used to allow the user to control the power at the output. Figure 2.8 shows the
overall PA circuit where the external DCO (digitally controlled Oscillator) can be
replaced with a VCO (Voltage Controlled Oscillator) running at 900MHz [15]. The
user inputs (Inputl thru Input8) will decide which transistors should be ON/ OFF.
When all transistors are ON, the PA delivers its most power, and when all transistors
are OFF, the PA should be turned off. Therefore to do this simple selection with the
DCO and the AND gate designed is just a simple static CMOS AND gate.

+5V

RFC

2
000 = M‘ X _I_ ll—‘ m ‘
Inputt: ) l CI

Input2 : ) l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

p—-=-ooo
Z
(I)

 

lnput8 .____. ) I

Figure 2.8. 8-bit Digitally Controlled Class E PA

 

Figure 2.8 was simulated with 8 different input levels using Cadence Virtuoso
to study the power at the output. Table 2.1 shows the results of these simulations.
Each simulation was conducted by itself. However, as the number of ON transistors
increases the contribution to the overall power of the new transistor added decreases,

which explains/proves that the circuit is actually nearing saturation. This can be

13

Table 2.1. Simulation Results for Output Power with Different Inputs.

Nunﬁmr of ON Transistors Output Power (dbm)
—163
—12.3
—9.01
—7.25
—6.01
—5.04
—4.25
—3.58
—3

 

 

 

 

 

DONOEUIAODNHC

 

clearly seen in ﬁgure 2.9.

 

 

 

 

 

 

Output Power (dbm)
do

I /
-12 L

-14 .. - __g A AL.
Number Of transistors

 

 

 

 

Figure 2.9. Output Power 8-bit Digitally Controlled Class E PA

2.2.4 Thermometer Decoder

To keep the size as minimal as possible and to keep the number of pins on the chip as
minimal as possible, a 3 to 8 decoder is implemented. The way this decoder operates
is different than the known standard decoders because we don’t just want to send the

PA one bit at a time. The reason behind that. is the way the PA functions; when one

14

Table 2.2. Thermometer Decoder INPUT /OUTPUT

 

 

3—Bit Input 8—Bit Output
000 00000001
001 00000011
010 00000111
011 00001111
100 00011111
101 00111111
110 01111111
111 11111111

 

 

 

 

transistor is ON that gives us one power level when two transistors are ON that gives
us two power levels. Therefore, we need a decoder which sends more than one bit at
a time to cause the PA to turn on more than one transistor simultaneously. So, the
best choice would be a thermometer decoder which can be summarized in table 2.2.
At any point there is at least one transistor switched ON. This guarantees that the
number of ON switches can be controlled precisely because there is no overlap. In
addition, because of the fact that the transistors are connected in parallel, the location
of the ON switch does not really matter so the order of bits in the Thermometer
Decoder is not crucial.

The design of this decoder was accomplished using VHDL and Cadence Silicon

Ensemble and it was optimized to use the minimal real-estate possible on the chip.
Figure 2.10 shows the VHDL code for this decoder.
The schematic generated using this VHDL code contains 9 gates as illustrated in
ﬁgure 2.11. The overall area of the decoder is about 73um2. The standard cell used
to generated this decoder is OSU Standard Cell Library v2.1[16] along with NCSU
CDK v1.5.1[17].

15

 

library IEEE;
use IEEE.stdﬁloch 1164.311;
use IEEE.Std‘loqic_unsigned.all;

entity LOGC3T08 in
port (SEL: in SID_LOblCAVRCI0R(x donate 0);
LOGC_OUT: out 8

TDiLOGlC-VECTUR(- downto 3));
end LOGC3T08:

architecture BEHAV of LOGC3TOB is

begin

WITH SEL SELECT LOGC‘OUT <=
' lil.." IHEN " 11",

"ll.lli.f" ”_'
liilll ﬂ" WHEN "i '"

"ll.ll”w " WHEN "[rn",

"I‘ ~"wazN'wI,

"In WHEN “ ,

"Mil ' wean OTHERS:
end BEHAV;

 

 

 

Figure 2.10. VHDL Code for Thermometer Decoder 3—to—8

LOGC_OUT

 

Figure 2.11. 3-to-8 Thermometer Decoder Schematic

2.2.5 Magnetic Force Actuator

The magnetic force actuator is simply a spiral inductor replacing the RL load in the
Class E PA described in the previous section. The design criterion for the magnetic
force actuator is creating a three metal layer inductor with the smallest dimensions
possible. This resulted in a 6x6um2 square with three layers as shown in ﬁgure 2.12.
The windings of the spiral inductor for each layer is not a full square, one of the sides
is misssing so that makes a three quarter square or a square with three sides. Again,

the inductor was designed using ASITIC and its inductance came out to be 20pH.

Figure 2.12. A 6 micron by 6 micron Actuator

If we unstack the metal layers in ﬁgure 2.12. The shape of the metal layers would
look like ﬁgure 2.13. Note, that Metal3 is one the left, Meta12 is in the middle, and
Metall is on the right. The access pin of this actuator is located on the upper right
corner of Metal3 layer and the exit pin is located on the upper left corner of Metall.

The layers are connected using the smallest sized VIAl and VIA2.

Figure 2.13. Unstacked Metal Layers for The Actuator

 

The actuator can be approximated with a Solenoid, consequently, when a current

17

passes through it the actuator can generate a uniform magnetic ﬁeld similar to the
one generated by a bar magnet. However, the direction of the magnetic ﬁeld will be
alternating because the output of the PA is AC. The 3D structure of the actuator

along with the direction of the current is previewed in ﬁgures 2.14 and 2.15.

Input

/- Metal3

  
       

\— Metall

VIA1

Output —'

Figure 2.14. 3D Angle View for The Magnetic Force Actuator

Metal3 ~\\\ /_ Input

MetalZ
VIA1 ,

Out ut —/"/ f
p Metall —/

Figure 2.15. 3D Side View for The Magnetic Force Actuator

The Class E PA will have the inductor shown in ﬁgure 2.14 as a load. So, all of
the current going into the load will be actually going through it, therefore, it will

generate a magnetic force coming out of in the center of the inductor depending on

18

the direction of the current. The magnetic ﬁeld can be approximated as in equation

(2.6)[18]

N
B = ”SiOQII (26)

where B is the magnetic ﬁeld, ,u permeability, N is the number of turns, L is the
length (in our case, the length is in the Z direction), and I is the current. The X and
Y dimensions of the spiral inductor are 6pm by 6pm however, the Z is not known.
The technolog ﬁle model used to design this spiral inductor in ASITIC states that
the thickness of Metall is 0.5um, Meta12 is 0.5/rm, and Metal3 is 0.8um. The distance
between M1 / M2 is 0.5um, and M2 / M3um is 0.5um as shown in ﬁgure 2.16, therefore,

The unknown Z is 2.8,um.

Overall Thickness 2.8um

 

‘ 1
a Metal3 thickness 0.8umj

x l M2M3 Spacing 0.5um
QT Metal2 thickness 0.5um

‘ '7 x... l M1M2 Spacing 0.5um
El Meta|1 thickness 0.5um j .

 

 

 

 

 

 

Figure 2.16. AMI 0.5 Metals and VIA Properties

Also, #5502 = 1/02551-02 = the reciprocal of the speed of light squared times the
permittivity of S 202, which equals to 3.18x10'7. Now N = 3 turns, L = 2.8um, and
I = 2'(t) and it is a function of time as seen in ﬁgure 2.17. The Frequency of the
waveform is 900MHz. Thus, the magnetic ﬁeld will be alternating and will have two
different directions per cycle. The maximum ﬁeld generated is exactly at peaks of
the current function which are located at g}; and ‘2 of the period where the magnitude
is approximately ISImA. This will give us a rough estimate of B which equals to
2.724x10'3 Tesla. However, to get a better magnetic ﬁeld off the chip we designated
an “UNPASSIVATED DIE” with a glass cut on top of the actuator (20pH inductor).

19

41 1 20h
Unu3( s )

 

 

 

 

 

 

 

 

 

_:.> 4
C’"
:3 4
C
C— i
__‘_'::.>
C; .4

 

‘ L iﬁn

 

 

 

 

 

 

 

 

 

/L@/PLUS

 

 

Slﬂn1,':

 

(v)

Figure 2.17. Current Passing Through The 20pH Inductor

20

2.3 Conclusion

An on-chip digitally controlled Class E Power Ampliﬁer with an on-chip spiral induc-
tor connected as load to actuate magnetic force is designed, simulated, and fabricated

using the AMI 0.5um technology. Figure 2.18 illustrates the ﬁnal schematic of the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ﬁnal circuit.
+5V
RFC
x ('3'2 L2 Out
DCO . M‘ _L II M "
__ C Actuator
Input1: 1
M2 '
lnput2 : ) l

2
oo

 

lnput8 .__. > I

Figure 2.18. Final Schematic for The Magnetic Force Actuator

 

The pad frame area is 1.5mm x 1.5mm, the actuator with all components consumed
535nm x 378nm area as shown in ﬁgure 2.19, so other three PAS were placed for test

purposes. The full layout has been included in Appendix A.

21

 

_ r,

H \H
\.
\..

an.“

m...

.

”Mu. I.

., i

x

 

 

 

 

ic Force Actuator

Figure 2.19. Final Layout for The Magnet

22

CHAPTER 3

Wireless Control

3.1 Introduction

This chapter covers the requirements, design, and results for the Wireless Control. As
discussed in the previous chapter, an 8-bit digitally controlled magnetic force actuator
was designed and fabricated, so, to utilize its small design and low power properties,
we used the Zigbee protocol to interface and control itwirelessly. Because Zigbee was
designed for low data rate, low power, and reduced resource requirements we chose it
over other protocols. The main goal is to create an arbitrary number of magnetic force
actuators to form a wireless sensor network. A single Coordinator (Controller) will
operate the sensor network with two methods: It can either send commands based
on a preprogrammed control algorithm or it can send commands using a manual user
input. The controlling commands are actually a 3-bit representation of the desired
magnetic force to be generated. The Microchip Zigbee ICDEM Z kit was adopted to

provide a platform for this wireless control.

23

3.2 Zigbee Protocol

3.2.1 Introduction

The Zigbee Protocol follows the IEEE 802.154 speciﬁcations for its Medium Access
Layer and Physical Layer. It uses the ISM radio bands; 868MHz, 915MHz, and
2.4GHz. For each band there exists a ﬁxed number of channels to operate on. The
bit rate is variable and it depends on the operational frequency. 868MHz can provide
up to 20kbps, 915MHz up to 40kbps, and ﬁnally 2.4GHz up to 250kbps [19]. These

numbers are summerized in table 3.1.

Table 3.1. Zigbee Operational Frequency, Bit Rate, and Number Of Channels

Frequency Bit Rate Channels
868M H z 20mm I
915MHz 40kbps 10

2.4GHz 250kbps 16

 

 

 

 

 

 

 

3.2.2 Zigbee devices

The IEEE 802.154 specifies two types of devices: FFD(Full Function Device) and
RFD (Reduced Function Device). The main difference between these two devices is
the power. For an FFD device it is assumed that it can offer all kinds of services while
the RFD can only support limited services. Also, because the FFD has to support
everything it must stay “awake” all the time and this requires a constant reliable
power source. On the other hand, the RFD will “sleep” when it is idle, therefore, it
will run on batteries or other low power sources.

The Zigbee protocol emphasizes these types differently depending on the applica-

tion. There are three device types stated in the Zigbee Protocol:

0 Coordinator: Modeled as a F FD. It forms the network, allocates addresses, and

keeps the binding table.

24

o Router: Modeled as a FFD. It extends the range and capacity of the network

and can have monitor or control functionality.

0 End Device: Modeled as a RFD (or even FFD). It has monitoring or controlling

functions.

In this work, the magnetic force actuator chips are considered as “End Devices” and
they operate as Reduced Function Devices to utilize their low power properties. The
actuators will not communicate with or control each other because that job is assigned
to the Coordinator (FFD) only. This even reduces the RFD requirements further.
Therefore, the network architecture used for this work is “Start Network Conﬁgura-
tion” were there exists only one Coordinator and several End Device nodes (magnetic
force actuators) as illustrated in ﬁgure 3.1. There are several other conﬁgurations
available in the Zigbee Protocol like “Cluster "ﬂee Topology” and “Mesh Network”.
By using the simplest model possible, we guarantee that the software implementa-
tion remains as minimal as possible, which reduces the overall overhead and required

microprocessing resources.

(< >) (t 1)
“\ I/
9.2.... /-/,’
MFA /./' MFA

 

 

 

 

 

é-bit abii
".\.. 'I'f
. If”!
_ (r )) f",
.// U \I'\
Coordinator “\.‘
/ (Broadcast) ‘
as? >54»;
,, A E) .,\\
6 'g ‘-
til 2‘
,/'l o a) ‘.\
(< r) .9) E (( ))
. ea
3 3,;
MFA l 3 MFA

 

[ ' MFk Magnetic Force Actuator Node

 

 

Figure 3.1. Magnetic Force Actuator in A Star Network Conﬁguration

25

Each node (FFD or RFD) has two addresses: a 16-bit network address which
is acquired once a node successfully joins a network and this address will be used
to communicate with the coordinator. The second address is a 64-bit MAC address
which is hardcoded and it has to be globally unique. This is analogous to IP (Internet
Protocol), where each device has an IP address and a MAC address, The MAC address
is hardcoded on the ethernet device for example while the IP address is obtained once
the device joins/ forms a network.

The Zigbee protocol uses the IEEE 80215.4 MAC which is 127 bytes long and it
contains a 16—bit CRC value for error checking. It can also contain ACK ﬂags for data
transfer acknowledgment. A packet with an ACK ﬂag requires the targeted node to
conﬁrm the reception of that packet, however, that only assures receptions but not
complete processing. A node might receive a packet but might not process it due to
some limitations. Therefore, this kind of conﬁrmation should be implemented in the
application layer. it is important to know that, with the ACK ﬂag the sender will
retransmit the packet for a ﬁxed number of tries then report failure.

The Coordinator can send packets to a speciﬁc (MFA) Magnetic Force Actuator
Node by providing its 16-bit destination address in the MAC layer header. This
method is called Unicast and it is not currently used but it can be easily enabled
to expand the functionality further. On the other hand, Broadcasting is used in this
work, the destination address in the MAC layer is replaced with the broadcast address
OXFFFF. Any MFA will be able to receive these packets because Zigbee implements a
passive acknowledgment feature where all broadcasting packets have to be rebroadcast
again by all nodes and if one node does not rebroadcast other neighboring nodes will
notice that it missed the broadcast packet and they will retransmit until the missing

node hears it or time out occurs.

26

3.2.3 Zigbee Stack

A free implementation of the Zigbee protocol is provided by Microchip. Their Zigbee
Stack v1.0—3.8 was used [2] In this work, there are some limitations in this implemen-
tations. However, to control the Magnetic Force Actuator nodes wirelessly, none of
these limitations affected us. Nevertheless, it provided extra functions which can be
used later to expand the functionality of this wireless control. Figure 3.2 shows the

Zigbee protocol stack architecture.

 

 

Applioaiiciﬁ‘suppon Subiawmps) 1
lﬁPS Message r‘ [ APS Security

 

, Management Management

 

HWE— New Lay'er "

Routing NW K NWK Security _ ‘ 7" '
, = Management 3r- Management 3" Management .-,—.- ,.
.__ 5.. i l. " V" WJK 7". , . ”‘4’.

SSP Interface

 

 

 

 

 

 

 

 

. m;
PHY (iEEE 80215.4) l

 

 

2.4GHz I I 868/915MHz I

 

 

Figure 3.2. Zigbee Protocol Stack Architecture [2]

27

3.3 Wireless Nodes

As discussed in the previous section Zigbee provides two device models FFD and RFD.
Each device has its own properties, thus, to take advantage of these properties and
to make them ﬁt with the requirements of this project the Controller (Coordinator)
will be represented as an FFD and it will be responsible for forming the actuation
network. On the other hand, the Magnetic Force Actuator nodes (MFAs) will be used

as RFD devices and will only listen to commands broadcasted by the Controller.

3.3.1 Controller (Coordinator)

The Controller is assumed to have “unlimited” resources. That is the power supply,
CPU resources, and memory. Thus, it will be responsible for organizing, commanding,
and controlling other MFAs. The main properties of this node can be described as

following:

o It remains awake all the time in order to assure that all MFA nodes are func-

tioning correctly.

It is aware of all possible MFA nodes that might join the network.

It initiates and manages the network for other MFA nodes.

It has all the preprogrammed algorithms for controlling other MFA nodes (au-

tomated mode).

It accepts manual user input commands for controlling other MFA nodes (man-

ual modes).

The Controller consists of four major parts: a microprocessor, 3 input switches,
three push buttons, and a transceiver. First, The microprocessor used is an 8-bit

PIC18LF4620 from Microchip. It processes user inputs (using interrupts shown in

28

ﬁgure 3.3 ), preprogrammed control algorithms, Zigbee stack, and interfacing with
the transceiver and other components on the board. Second, the 3 input switches
SW2, SW1, and SWO are used as manual user input to control the MFA nodes,
these switches are connected RDO, RDl, and RD2 on the microprocessor. The three
switches represent the desired magnetic force actuated each switch triggers one bit,
so, we have three bits and 23 possible numbers. Therefore, we can command the MFA
nodes to produce eight different levels by using these switches as shown in table 3.2

and this is similar to what the Class E Ampliﬁer had in table 2.2.

 

void UserInterruptHandler(void)

l

// I: this on interrupt-on-chonge interrupt?

#ifdef IMA_CONTROLLER

if ( INTCONbits.RBIF =8 1 )

l
// Record which button was pressed so the moinO loop can
//hmdeﬂ

if (AUTO_SWITCH == SWITCH_PRBSSED)

{
myStatusFlags.bits.autoActuationStarted = TRUE:
myStatusFlags.bits.autoActuationSwitch = TRUE:
ConsolePutROMString( (ROM char *)"Start Automatic Actuation...\r\n" );

1f (SEND_SWITCH == SWITCH_PRESSED)
{
myStatusFlags.bits.userInputSwitch = TRUE:
ConsolePutROMString( (ROM char *)"Start User—Input Actuation...\r\n" );

l
// Disable further RBI? until we process it
INTCONbits.RBIE 8 0;

// Clear unis-match condition and reset the interrupt flag
LATB 8 PORTB;

INTCONbitS.RBIF = O;

l
#endif

 

 

 

Figure 3.3. Coordinator Interrupt Routine to Start Actuation

Third, the three push buttons MCLR, S2, and S3 are used for initialization and

triggering. The MCLR push button will reset the microprocessor to its initial state,

29

Table 3.2. Manual User Input Using Switches

 

 

 

SW2 SWISWO Magnetic Force Actuated

000 SmallestForce

001 Forcel

010 Force2

011 F0rce3

100 F orce4

101 F 0rce5

110 ' F0rce6

111 LargestForce

 

 

 

S2 is used to trigger the microprocessor to sample the inputs on SW 0, SW 1, and SW 2

illustrated in ﬁgure 3.4 and then

 

if ( myStatusFlags.bit3.userInputSwitch )
{
myStatusFlags.bits.userInputSwitch = FALSE;

ZigBeeBlocka();

TxBufferthData++] = APL_FRAME_TYPB_KVP I 1; //KVElimmumnMn
TxBuffer[TxData++] = APLGetTransId();

TxBufferlTxData++l = APL_FRAME_COMMAND_SET I (APL_FRAME_DATA_TYPE_UINTB << 4);
TxBufferlTxData++J = OnOffSRC_OnOff 6 OxFF; //AﬂﬂhﬂeIDLSB
TxBufferlTxData++J = (OnOffSRC_OnOff >> 8) G OXFTV l/AﬁﬂhﬂeHJMSB
SR_DATA = O:

SR_DATA = (USER_I02 5 0x01):

SR_DATA = SR_DATA << 1;

SR_DATA = (SR_DATA E 0x02) | (USER_IOl & 0x01);

SR_DATA = SR_DATA << 1;

SR_DATA = (SR_DATA 5 0x06) | (USER_IOO & 0x01);

TxBufferITxData++J = SR_DATA;

PrintChar( SR_DATA ):
ConsolePutROMString( (ROM char *)" SR_DATA being sent.\r\n" );

 

 

 

Figure 3.4. Coordinator User Input Manual Actuation

broadcast the resulting three bits to the MFA nodes, and ﬁnally the S3 push button
triggers the microprocessor to start the automated mode were the preprogrammed
algorithm will be executed and then broadcasted to the MFA nodes as shown in ﬁgure
3.5. The aglorithm demonstrated in the ﬁgure is just is a simple counter which will
command each MFA node to actuate slowly from lowest force to highest then stOp.
The fourth and last part is the transceiver. The Zigbee stack supports two

transceivers Chipcon CC2420 and Microchip MRF24J40. In this work the Chipcon

30

 

if (myStatusFlags.bits.autoActuationSwitch)

{
count++;
myStatusFlags.bits.autoActuationSwitch = FALSE:
ZigBeeBlocka();

TxBuffer [TxData++] = APL_FRAME_TYPE_KVP | 1; // KVP,1tr-ansoction
TxBufferlTxData++] = APLGetTransId();

TxBuffer[TxData++] a APL_FRAME_COMMAND_SET I (APL_FRAME_DATA_TYPE_UINT8 << 4);
TxBufferlTxData++J = OnOffSRC_OnOff 8 OXFF; //AﬂﬁhneIDL$8
TxBuffer[TxData++] = (OnOffSRC_OnOff >> 8) & OxFF} IlknﬂhneHDMSB

SR_DATA = count-1;
TxBufferlTxData++l = SR_DATA;

PrintChar( SR_DATA );
ConsolePutROMString( (ROM char *)" SR_DATA being sent.\r\n" );

 

 

 

Figure 3.5. Coordinator Preprogrammed Auto Actuation

CC2420 transceiver is used and is controlled using the SPI interface. The Controller
node will never switch off its transceiever and it will always book keep other nodes

on the network.

3.3.2 End Devices (MFA Nodes)

The End Device is assumed to have “limited” resources. That is the power supply,
CPU resources, and memory. Thus, its job is to listen/poll for certain commands
from Controller then sleep to save power the main properties of these nodes are the

following:
o It should sleep whenever it is idle.
o It is aware of other MFA nodes but it does not manage them.

0 It does not try to perform its own network. It can only join pre—existing net-

works.

0 There are no preprograms installed on it and its code remains as minimal as

possible.

31

o It accepts broadcasted commands from the Controller only.

The major components of such nodes are: ﬁrst the microprocessor were the Zigbee
stack and minimal MFA functionality are stored, second, the transceiver similar to
that described in the Coordinator. However, it must be noted that the current Zigbee
stack is non-beacon which means that each MFA node should poll for data when it
wakes up from sleep. There are three output pins on the microprocessor which are
connected directly to the Magnetic Force Actuator Chip. Speciﬁcally, these three bits
will be connected to the 82, S1, and S0 on the 3-8 thermometer decoder which will be
later translated to different magnetic force using the Class E PA. For example, ﬁgure
3.6 shows when a data packet is received, it is processed in the RFD using a simple
switch statement to ﬁgure out which one of the eight input levels was requested.
The ﬁrst three cases are shown below. It is also possible as previously discussed, to

conﬁrm with the Controller that this packet was successfully received.

 

switch (data)
{

case DATA_ZERO:
ConsolePutROMString( (ROM char *)" Received ZERO.\r\n" );
USER_IOO 0;
USER_IOI 0;
USER_IOZ 0;
TxBuffer[TxData++] = SUCCESS;
break;

case DATA_ONE:
ConsolePutROMString( (ROM char *)" Received ONE.\r\n" );
USER_IOO 1;
USER_IOl 0;
USER_IOZ 0;
TxBuffer[TxData++] = SUCCESS;
break;

case DATA_TWO:
ConsolePutROMString( (ROM char *)” Received TWO.\r\n" );
USER_IOO 0;
USER_IOI l;
USER_102 0;
TxBuffer[TxData++] = SUCCESS;
break;

 

 

 

Figure 3.6. RFD Node Receiving 3—Bits That Represent. 8 Levels

32

3.4 Conclusion

Using the Zigbee protocol and the Microchip ICDEM Z kit. We provided a reliable
low power wireless control for the Magnetic Force Actuator. Three bits represent
eight different force levels which can be inputted or preprogrammed on a Controller
and then sent wirelessly to a network of Magnetic Force Actuating nodes, where each
node will perform the command it receives irrespectively and regardless of the other

nodes in the network.

33

CHAPTER 4

Atomic Force Microscope

4.1 Introduction

Atomic Force Microscopy was invented in 1986, which was based on Scanning Tim-
neling Microscopy 1981. AFM gives us the ability to get high resolution images of
surfaces (down to the nano scale). It can display features as small as an atomic lattice.
With this high resolution, scientists can even come closer to the atom more than what
they would have ever thought. AFM has three major components: a laster beam, a

photodetector, and a cantilever. The basic operation of an AFM is as follows:

o The cantilever which has a tiny tip (probe) approaches a surface to be scanned.

o A laser beam will be directed on the cantilever’s back surface while it is moving

over the surface.

0 The tiny tip interacts with the atoms on the surface causing the cantilever to
deﬂect upward or downward and that results in a change in the laser beam

reﬂection angle.

0 The photodetector records the change in reﬂected laser beam and relates it with

the deﬂection as seen in ﬁgure 4.1.

The recorded results can be represeted as a surface topography image with a very

high resolution. However, imaging the surface is not the only application AFM can

34

be used in, other useful calculations can be acquired from the interaction between the
surface and the tip. For example, if the surface and the tip had magnetic properties
then a relation can be established between them to represent the attraction/repeltion

forces.

Detector and
Feedback

Electronics

   
  
  
  

Photodiode

 

Sample Surface Cantilever & Tip

PZT Scanner

Figure 4.1. Atomic Force Microscope [3]

There are two different types of imaging modes in AF M, static (contact) mode
and dynamic (non—contact) mode. In the static mode, the controller inside the AF M
tries to keep the force between the tip and the surface constant by adjusting the the
Z axis (the distance between the tip and the surface), so, the change in the Z axis
will be recorded and analyzed. In the dynamic mode the cantilever will be oscillating
far away from the surface at a certain resonance frequency. Therefore, the amplitude
of oscillation, phase and frequency are monitored by the AFM for any changes which

then describes interaction between the tip and the surface.

35

4.2 Magnetic Force Measurements

4.2. 1 Introduction

To be able to record the output of the Magnetic Force Actuators discussed in the
previous chapters, we have decided to use AFM because it is better than other tech-

nologies [20] as listed in the following table 4.1:

Table 4.1. Comparison Between AFM and Other Technologies

 

[ jAFM jTEM [SEM [Optical |
U on C C SIlIIl 11H}

cost x - - -
v1ronmen air, u1 , vac- vacuum vacuum air,

uum, special

 

BS
68

 

 

We used the basic N anosurf EasyScan 2 AFM[21] which provides static (contact)
mode. Utilizing the features in the static mode and using it along with a magnetized
cantilever tip, we were able to analyze some properties of the Magnetic Force Actu-
ator, which are discussed in the next sections. The normal operation of the static
mode requires the tip to “touch” the surface, however, in this application (using the
magnetized tip) approaching the tip close to the surface will damage it. Consequently,

We decided to use the static mode in a special way which will be explained later.

4.2.2 Measurements

The setup for this measurement required the use of all three components of this thesis;
the fabricated chip, the wireless control, and AFM. The Fabricated chip was placed
under the AFM scan head, however, the packaging of the chip made this step diﬂicult.
Because the chip could not easily ﬁt under the scanning head special mounting tools

and wiring were designed to allow the scan head to move freely on top of the MFA.

36

Once the chip was installed under the scanning head, it was tested to make sure
it is functional. The test involved using the Controller to send certain commands and
observe the current consumption increases when a higher power was requested. Also,
another test was to make sure that the output frequency matches the desired one.
After seeming that, the circuit was turned off. A magnetically doped cantilever was
then magnetized using a strong magnet, then it was installed on the scanning head.
The next step is to switch on the MFA and send it a 111 input using the manual
user input mode on the Controller. By doing so, we make sure that the tip will not
collapse into the chip because the scan head controller will notify us as soon as there
is an interaction between the two, thus, by setting on maximum force level we will be
able to detect such interaction early on.

Using the naked eye by looking through the magniﬁer on the scanning head, the
cantilever is brought closer to the chip by moving the leveling the screws downwards.
It is important to keep the head leveled to get accurate measurements. After the
manual approach by using screws is completed, an automated approach using the Z
controller in the software of the EasyScan2 is applied. The Z controller is a standard
PID controller to keep the tip at a certain distance from the chip. The distance
between the tip and the chip is represented as “Set Force = 20nN” which means that
the tip will come closer to the surface (in the Z direction) until it detects a force with
a magnitude of 20nN.

Once the automatic approach is done and the tip is at a reference point Z = ZO
(where the force exerted is 20nN+, we freeze the Z controller, so, it does not move
the cantilever anymore. By freezing the Z controller we guarantee that the cantilever
remains at the same position (Z0) in all experiments. Now since we used the maximum
possible force generated by the MFA, we will not be able to detect the smallest forces.
Therefore, the tip is brought closer to the sample while keeping a safe range between

the tip and surface. The reference point now is Z = Z1 and this will be the ﬁnal Z

37

which we will use in all measurements. The Z controller is still frozen, we just brought
the tip closer so we can detect smaller forces for the next measurements.

Figure 4.2 shows the ﬁrst spectroscopy measurement when the user input is 111,
it should produce the largest force on the tip. There are 256 data points during this
measurement, each data point is an average of 8 samples which were taking separately.
There was a ﬁve second period between each sample. So, in total this measurement
took 40 seconds to complete, therefore, it contains 2048 measurements. All these
points were averaged. Therefore, the total average deﬂection for user input 111 is

5.21523x10-08m.

 

Deﬂection - Spec forward

 

 

 

 

 

 

 

 

 

E __
C
I!)
S‘.‘
0)
U)
s S
CU , I—
'o __ c
g .2
It s
8
E
in I ..
(\l‘ 0 m Z-AXlS 0 m

 

Figure 4.2. Tip Deﬂection at 111

While keeping the same reference Z value (keeping the tip at a distance Z = Zl)
the user input will be changed to 110 and the deﬂection will be recorded again. The
same method is used 256 points with 8 averages and ﬁve seconds in between. The
deﬂection produced for user input 110 is 4.91725x10‘08m and shown in ﬁgure 4.2.
The same measurements are carried for the rest of the levels 101,100,011,010,001, 000,
and ﬁnally when the chip is switched off. The ﬁnal results as shown in table 4.2 and
ﬁgure 4.4. ,

The spring constant K for the magnetic cantilever was provided as a range between

1 to 5 from the manufacturer [22], so, we will represent the corresponding forces as a

38

 

Deﬂection - Spec forward

 

124nm
H

1

Raw data

 

l .
Deﬂection range

...................................................................

 

 

 

 

 

 

 

-22.6nm

 

Figure 4.3. Tip Deﬂection at 110

Table 4.2. Different User Inputs with their Corresponding Deﬂections

 

 

User Input Tip Deﬂection

OFF 2.41301E—10
000 2.79E—08
001 3.29E-08
010 3.73E-08
011 4.00E—08
100 4.44015E—08
101 4.79164E—08
110 4.91725E—08
111 5.21523E-08

 

 

 

 

range as well. The actual K will be sent to us in the near future. Table 4.3 shows
three possible values of K and their corresponding force. A graph of that is shown in
ﬁgure 4.5.

As previously discussed and shown in ﬁgure 2.17, the current generated by the Class
E PA will be running at 900MHz, it will cause the actuator to produce a changing
(alternating) magnetic ﬁeld at that frequency. However, the cantilever’s resonance
frequency is only 60KHz so the cantilever itself won’t be able to achieve and oscillate
at such high frequencies, so, it will act as a ﬁlter giving us an average RMS value.
Therefore, the results we are seeing in table 4.5 represent the averaged force after

being ﬁltered out by the cantilever.

39

6.00E-08

5.00E-08

3.00E-08

Deﬂection

2.00E-08

1.005433

 

0.00E+OO

NONE 000 001 010 011 100 101 110 111
Input

Figure 4.4. Tip Deﬂection at All Input Levels

Table 4.3. Different User Inputs with their Corresponding Forces

 

 

—In—put orce at K=1 Force at K=3.0 Force2 at K=5
OFF 2.4lE-10 6.0325251” 1.2065509
000 2 . 79E—08 8.38E—08E—10 1.40E—07
001 3.29E—08 8.23E—08 1.65E—07
010 3.73E-08 1.12E—07 1.87E—07
011 4.00E—08 1.20E—07 2.00E—07
100 4.44E—08 1.33E—07 2.22E—07
101 4.79E-08 1.44E—07 2.40E—07
110 4.92E—08 1.48E—07 2.46E—07
111 5.22E—08 1.56E-07 2.61E—07

 

 

 

 

 

 

4.2.3 Conclusion

Using the AFM technology, we were able to compute the force actuated from the
MFA at a certain distance in the Z direction. A Magnetically doped cantilever was
used and placed on top of the 20pH inductor, for an example see ﬁgure 4.6. Arrow 1
points at the base of the cantilever. Arrow 2 points at the cantilever itself. Note that
the cantilever is actually covering the small 20pH inductor beneath it. The range of
the force generated out of the MFA, at the same Z distance, is from 2.41x10‘10N to

5.22x10‘08N with a cantilever spring constant K=1 and 1.40x10—9N to 2.61x10‘07N

40

 

3.00E-07

2.50E-07

 

 

2.00E-07

1 .50E-07

 

 

Force

1 .00E—07

 

-0- Force K=3.0
+ Force K=1.0
Force K=5.0

 

 

5.00E-08

 

 

 

0.00E+00 ""'"'_" r r r r

 

 

 

NONE 000 001 010

011
Input

100

101

110

111

Figure 4.5. Force Exerted On Tip at All Input Levels At Distance Zl

with a cantilever spring constant K=5.

41

 

4‘36’M'M'

C l

L‘DLL (are.

i
U
a
b
'-

'i
0‘1‘
9

“I;

 

’a '1’; '4 f4 5. E. ii-
. . ' ' r

Figure 4.6. Magnetic Cantilever on Top Of MFA

42

CHAPTER 5

Conclusions

5.1 Results

We were able to design, simulate, and test a wirelessly controlled magnetic force
actuator as illustrated in ﬁgure 5.1(a) and 5.1(b).
I, L "b o '5 o L t o c

a:

a
v
‘

 

(3.) MFA Test Layout (b) MFA Fabricated Test Chip

Figure 5.1. Magnetic Force Actuator Chip

It can produce eight different levels of magnetic forces as shown in ﬁgure 5.2. These

forces were measured using a magnetically doped cantilever shown in ﬁgure 5.3

43

3.00E-07 .. 7f

2.50E-07 t—l

zoos-07 '

 

 

 

-°— Force K=3.0
1.505437 /’

+Force K=1.0
/ Force K=5.0
1.00507 //
5.00E-08 ‘ 7 ,., - w:—
//I r7,wﬂ

0.00300 9 ' ._ . . . .
NONE 000 001 010 011 100 101 110 111
Input

 

Force

 

 

 

 

Figure 5.2. Force Exerted On Tip at All Input Levels At Distance Z1

i""l,._‘ n [\I‘ '

0 3.. .
\LLLL;4¢..

‘0

b
..
.
U
a

'1’
O

II I; .
1

 

'a‘ ’a io. up i. i. i-'
i \

Figure 5.3. Magnetic Cantilever on Top Of MFA

The overall work can be summerized in the following ﬁgure 5.4. A wireless network of

Magnetic Force Actuators which can be controlled using a Controller (Coordinator)

to produce eight different output levels using a preprogramed algorithm or a manual

user input.

JDJJJDJJZDDJJDDED

 

333333

cameos

 

0 87654321

 

9 8 7 6 5 4 3 2 1
222222222
555:2
826.". m

. «3.0 .355“

Es

 

 

 

 

 

 

 

 

 

33333353333333333333

 

 

 

 

 

 

 

 

 

 

   

 

 

 

09878....ou4m21098ﬁ65 ﬁﬂﬂ
cameos
.9265 Kim
. : 5. .. ._oE_QE< Louoooo
. . .1 526a m 86E
. ._._ $20 .955 . oéoﬁ
. x ebé we;
. 1 2 3158 7. 8 9 0
7 8 9 m 1 1 1 1- 1 1 4| 2
EECﬁNI/‘JCEECEEECEEE
sE<

..0>m _ cm

 

 

 

storm:
oonmﬁ

a e

Se.

r

=m...m

  
 

lonuoo pewurerﬁord

Ionuog Jasn

 

as.
Esta
ﬁ Kt...

   

gm.

M68399
aoymcﬁcooo

Q av

an

     

Figure 5.4. Thesis Block Diagram

45

5.2 Future Work

For future development it would be desired to have a more precise model for on-
chip inductors and capacitors. After testing the fabricated chip it appeared that the
output frequency was not tuned correctly. This is due to a mismatch between the
values used in the simulations and the result values from the fabrication process, the
desired frequency was supposed to be 900MHz however, using a frequency counter at
the output pin; the tested frequency was 600MHz. Also, It would be better to design
on—chip test components to be able to measure and test the Class E Power Ampliﬁer
internally instead of using external tools because currently available external testing
devices could not precisely measure the output power of the PAE. Additionally, an
AFM scanner with a movable base that can hold the circuit and move it around
is needed because it would place the cantilever tip precisely on top of the target,
instead of using manual hand maneuvers. Locating the 20pH under the tip was the
most diﬂicult part: ﬁrst, the packaging process after fabricating the chip places the
bonding wires upwards (coming out of the pad frame in the Z direction), so this made
it difficult for the cantilever to reach the surface of the chip, which might break the
bond wires or the cantilever itself, therefore, it would be better to have a different
type of packaging which places the bond wires on the sides instead of the top. Second,
the current method used to place the tip exactly on top of the 6ux6u uses the naked
eye as a guide, consequently other more accurate methods should be used to precisely
place the tip. Because of these two reasons it was difﬁcult to repeat the measurements
and get the same exact values of the force for different setups. As a result, future work

should have a different layout and packaging to easily locate and scan the actuator.

46

APPENDIX A

Magnetic Force Actuator Test Chip

.-

a.
a.
m
x
“.
x
w
.n
w
\.
v.
m
x
w
a.
_\
n,
.t
x
w
x.

\\\\\\‘ \\\\ \\'~\x\\\~ .\ . v we \\\
l
‘\\\\\\\\\\\s\\\\\\\\\\\\\\ \\\\I\‘-\\\§\'\\\\\i»\\\-,\\\\\\\\\

. \\~.\\\\\\\\\~.\\x\\\\\\~s\\ .\\\\\\\\\\~ ~\\\\ .\.\tn\\w\\~w

\\\\ llxxxxx §§§§§x§ \\\\\\\§ »\\\\§ §§§§<§§l§§>

\\\ tit. 2x35. «\Exaixxx ..\\\\\ {xxx .Sxxxxxxx.

 

{stktix \\\§\§\~\ \xxtxx \\\\§ ~\\§\\\§\§\\\ . .
ixttxxt \\\\.\\¢\\\\§§\\\\ \ stxx \s\\\\\\ C \xxx,

\\\§\\\\§\\.§\\\§t \\\\\\\\\\\\\n\\\\\\\\\\\\\\\~\\\§\\\\.
.\\\\.:\\§ .<\\\\\\?§\\\\.5 \\ xx§~§ \\<\\\\ \\\\\)C

‘\\\\'\

 

. \
{\xS

.xs“\u

 

Figure A.1. MFA Test Chip Layout

47

 

Figure A.2. MFA Fabricated Test Chip

48

Test Pins + V00! 0 D 52
U C SO 1
T
/

Test Pins [VDD ]

ECND ]

\ Test Pins
‘\ ﬂ . .AR

DEREK “REL
1111111111

//

Test Pins + GND!

 

/

 

 

\\\ \ l I / [I
trauma Uuuuu
Imam]: marina

 

 

 

 

 

\

\

\

 

 

 

Figure A.3. MFA Pin Diagram

49

BIBLIOGRAPHY

[1] N. O. Sokal and A. D. Sokal, “Class E - A new class of high—efﬁciency tuned
single-ended switching power ampliﬁers,” IEEE Journal of Solid-State Circuits,
vol. 10, pp. 167—176, Jun. 1975.

[2] Microchip. (2007, May) Zigbee stack. [Online]. Available: http://www.
microchip.com/zigbee/

[3] Wikipedia. (2007, May) Atomic force microscope. [Online]. Available:
http: //en.wikipedia.org/wiki/AFM

[4] J. M. Rabaey, A. Chandrakasan, and B. Nikolic, Digital Integrated Circuits A
Design Perspective. Prentice—Hall, 2006.

[5] S. .1. Ball, C. Folsom, and A. B. McLean, “A compact nano-positioning stage
with high vibrational eigenfrequencies,” 2005.

[6] H. Lin, B. Lu, Y. Ding, Y. Tang, and D. Li, “A motor-piezo actuator for nano-
scale positioning based on dual servo loop and nonlinearity compensation,” 2003.

[7] MOSIS. (2007, May) Ncsu cdk. [Online]. Available: http://www.mosis.org/
products / fab / vendors / amis / c5 /

[8] F. T. Abu-Nimeh and F. M. Salem, “Digital class e power ampliﬁer for the rf
band,” Proceedings of the IEEE International Conference on Mechatronics and
Automation (ICMA ’06), 2006.

[9] B. Razavi, RF Microelectronics, lst ed. Upper Saddle River, NJ, USA: Prentice
Hall, Inc., 1997.

[10] S. C. Cripps, RF Power Ampliﬁers for Wireless Communications, 2nd ed. Nor-
wood, MA, USA: Artech House, Inc., 2006.

[11] T. H. Lee, The Design of CMOS Radio-Frequency Integrated Circuits, 2nd ed.
Cambridge University Press, 2003.

[12] A. M. Niknejad. (2007, May) Asitic. [Online]. Available: http://rﬁc.eecs.
berkeleyedu / ~niknej ad / asitic.html

50

[13] D. K. Choi and S. 1. Long, “High efﬁciency class e ampliﬁers for mobile and base
station applications.”

[14] P. Alinikula, K. Choi, and S. 1. Long, “Design of class e power ampliﬁer with
nonlinear parasitic outputcapacitance,” Circuits and Systems 11: Analog and
Digital Signal Processing, IEEE Transactions, 1999.

[15] P. Cruise, C.-M. Hung, R. B. Staszewski, O. Eliezer, S. Rezeq, K. Maggio, and
D. Leipold, “A digital-to-rf-amplitude converter for gsm/gprs/edge in 90-nm
digital cmos,” 2005.

[16] O. S. University. (2007, May) Osu standard cell library. [Online]. Available:
http: / / vcag.ecen.okstate.edu/ pro j ects / scells/

[17] N. C. S. University. (2007, May) Ncsu cdk. [Online]. Available: http:
//www.eda.ncsu.edu/

[18] HyperPhysics. (2007, May) Solenoid magnetic ﬁeld calculation. [Online].
Available: http: / /hyperphysics.phy- astr. gsu.edu / hbase / magnetic / solenoid .html

[19] hasse.nl. (2007 , May) A technical overview of wireless technology. [Online].
Available: http://zigbee.hasse.nl/

[20] Agilent. (2007, May) What is atomic force microscope. [Online]. Available:
http: //www.molec.com/whatjsafmhtml

[21] Nanosurf. (2007, May) Nanosurf eastscan2. [Online]. Available: http:
/ / www.nanoscience.com / products / easyScan2 .html

[22] AppNano. (2007, May) Magt data sheet for magnetic force microscopy probes.
[Online]. Available: http: / /www.appnano.com /pdf/ Datasheet-MAGT.pdf

51

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

rngwgng