journal of advanced mechanical design, systems, and ... · journal of advanced mechanical design,...

Journal of Advanced Mechanical Design, Systems, and

Manufacturing

Vol. 4, No. 6, 2010

1119

Feasibility of Air Levitated Surface Stage for Lithography Tool*

Keiichi TANAKA** **Department of LSI Process Tool Development, Core Technology Center, Nikon Corporation,

471, Nagaodai-cho, Sakae-ku, Yokohama-city, Kanagawa 244-8533, Japan E-mail: [email protected]

Abstract The application of light-weight drive technology into the lithography stage has been the current state of art because of minimization of power loss. The purpose of this article is to point out the so-called, "surface stage" which is composed of Lorentz forced 3 DOF (Degree Of Freedom) planar motor (x, y and theta z), air levitation (bearing) system and motor cooling system, is the most balanced concept for the next generation lithography through the verification of each component by manufacturing simple parts and test stand. This paper presents the design method and procedure, and experimental results of the air levitated surface stage which was conducted several years ago, however the author is convinced that the results are enough to adapt various developments of precision machining tool.

Key words: Lithography Tool, Wafer Stage, Linear Motor, Planar Motor, Air Bearing, Motor Cooling, Magnetic Levitation

1. Introduction

Global warming has become a serious worldwide problem recently and there are two possible solutions to the problem as follows, 1) adoption of higher efficient system (stable system); and 2) promotion of 3 R (Reduce, Reuse and Recycle). The author, in parallel, has investigated the possibility of using 3R in semiconductor manufacturing tools in terms of costs and has found that the recycling cost of engineering ceramics is much higher than conventional metals. For this reason and others, some ideas of 6 DOF MAGLEV (Magnetic Levitated) stage which is basically composed of certain kinds of metals, have been studied. However, it is generally hard to eliminate the power consumption for this stage even if a zero-power concept(1) is adopted. Currently, in semiconductor lithography, the technologies of both ArF(Argon-Fluorine Excited dimer laser) immersion lithography with a multi-exposure system and leading-edge EUVL (Extreme Ultra Violet Lithography) have been extremely accelerated in order to realize the following top three indexes, "smaller and thinner patterns of a few dozen nanometer on wafer devices", "increment of throughput" and "reduction of footprint". Since these targets should be identified, materials of higher stiffness ratio(=stiffness/density) and elements of a higher power ratio(=force constant/mass) are the common benchmarks in not only this field but also in every transportation system from the viewpoint of controlling power consumption.

Some drastic innovations have been typically invented at the stagnation of economic and technical competition. In the early days, a linear drive system made up of linear motor and air bearing was generated in place of a non-linear drive system of rotation motor and rack and pinion unit to implement a higher bandwidth of control by eliminating non-linear

*Received 29 Mar., 2010 (No. 10-0132) [DOI: 10.1299/jamdsm.4.1119]

Copyright © 2010 by JSME

Journal of Advanced Mechanical Design,Systems, and Manufacturing

Vol. 4, No. 6, 2010

1120

parts, for instance friction and backlash. In general, it is said that innovation brought about a de facto standard of x and y stage design in a widespread industrial field, especially lithography tool, and was considered as the kind of common platform which is mostly effective for cost reduction. Recently, some MAGLEV 6 DOF concepts have been very actively investigated(2)(3)(4), however both the initial cost of R&D, including manufacturing and debugging cost, and the running cost of power consumption may not be offset no matter how effective this vacuum compatible concept is. There are many significant indexes to evaluate the lithography tool, indicated in followings; 1)preciseness, 2)throughput, 3)footprint, 4)cost(initial and running), 5)date of delivery , 6)tool weight, 7)facility(gases, electricity) and 8)environmental compatibility. The interaction between indexes largely exists; however, indexes for 1) to 3) should be comparatively independent of each other and should be fixed at the beginning of development. Furthermore, 4) and 5) as well as the others are relevant to the concept of each component of equipment. The MAGLEV 6 DOF "reticle" stage was already developed by the author(5)(6), obtaining good results because of the long scanning traveling and short orthogonal adjusting axis. On the other hand, it is easily shown that the development of MAGLEV 6 DOF "wafer" stage implies wasting a lot of time and resource for the large planar compensation. For example, the decoupling of the axes at x and y long traveling plane must be diligently overcome. Therefore the air-levitated 3 DOF wafer stage is focused on in this paper to reduce 4) cost(initial and running) and 5) date of delivery. This is on condition that an additional 6 DOF fine stage is mounted on the 3 DOF stage and a differential vacuum pumping system will be adopted for vacuum lithography tool(5). Some excellent investigations of the stage possibility have been already performed (7)(8)(9)(10); however it seems premature for a practice use in a lithography tool. In this paper, the concept of surface stage is introduced at first, then the procedure of design is indicated exactly.

2. Concept, details and analytical verification of design

2.1 Planar Motor The schematic diagram of surface stage is shown in Fig. 1 and is basically composed of

three unit, planar motor, air bearing and motor cooling system. The position of 6 DOF fine stage is measured and controlled by laser interferometers and the position of 3 DOF surface coarse stage is controlled by using the above interferometers and XY-2D encoder or gap sensors between fine and coarse stage that is not shown in Fig. 1.

The main specifications of surface stage is as follows: 1) acceleration(X & Y):2 G peak (recently updated to >5 G), 2) velocity(X&Y):0.5 m/s, 3) positioning accuracy:1 µm, 4) bandwidth of control(X,Y & θz):50 Hz (mechanical stiffness should be typically four times higher than the bandwidth) and 5) temperature rise of the moving stage:<0.1 degC.

Fig.1 Schematic diagram of air levitated 3 DOF surface stage with 6 DOF fine stage. The surface stage is

levitated on the guide surface by air injection underneath of stage. Both fine stage and surface stage of coarse stage

are measured independently, this figure shows the fine stage is controlled by laser interferometer of 0.15 nm

resolution.

3 dof Surface stage(Coarse stage):Target of this article

Fine stage

WaferOpticalsensors

Laser interferomters

Guide surface(Air bearing guide)


Vol. 4, No. 6, 2010

1121

Fig.2 Configuration of planar motor. Squared magnets attached to the moving metal plate (back yoke) are

arranged on every other pitch of magnet width in x and y plane. On the other hand, squared armature coils whose

width is coincident with the three times of magnet width, are mounted adjacent on the stationary plate (coil yoke).

Commonly, the motor is roughly divided into three driving methods, "Lorentz", "Pulse(Stepping)" and "Induction" without hybrid of them, and a Sawyer motor is very common as Pulse type and is developed at many organizations for its higher motor efficiency(9). However, high-band vibration and noise is increased in proportion to a rise of stage velocity, therefore some compensation methods of positioning control is required to achieve a higher accuracy. Meanwhile, there are generally two types of motor, moving magnet and moving coil, an advantage of the former is to be able to ignore tubes and wires of moving stage, on the other hand, the disadvantage is the need to permit a pitching and rolling torque because a driving force is generated at coils for the case of Lorentz type motor. However cooling of fixed coil is much easier and limit of temperature rise can be relaxed.

The concept of planar motor which is one component of surface stage is two dimensional array of moving magnet and configuration of two phase armature coil that corresponds to the magnet array. Figure 2 shows a concept of planar motor in which the magnets are aligned by every other pitch and an opposite polar on the gird of the plane. As shown in Fig. 2, since two phase coils are excited correspond to the position of magnet array, the total x and y force of the moving part is obviously constant.

Assuming that a) an origin of coordinate system (x,y) which is fixed on stationary coils is the center of one coil and b) a magnetic field at the coils is sinusoidal, the distribution of z component of magnetic flux density at magnetic field synthesized from multi magnets is represented as following. Bz(x,y)=Bz0・sin(x/Mw・π/2)・sin(y/Mw・π/2) (1) Here, Bz0 is a peak value of flux density, (x,y) is the location on the magnet array and Mw is a width of magnet. A width of coil, Cw is obviously three times of Mw. In addition, it is clear that one set of 2 x 2 coils (four coils) focused on detail area C in Fig.2 is magnetically repeated in the plane, so the z component of magnetic flux at active region of coil that contributed to the force for each x and y direction is expressed as, Bxi,j(dx,dy)=Bz0・sin(dx/Mw・π/2)・cos(dy/Mw・π/2) (2) Byi,j(dx,dy)=Bz0・cos(dx/Mw・π/2)・sin(dy/Mw・π/2) (3)

Detai Area B

W

SN

U

SN

NS

X

Magnetic flux circuit

Exciting current to generatemagnetic force

Magnetic force

Section AA

Permanent magnet

Back yoke

Armature coil

Coil yoke

Magnetic forceN

S

S

N

S

N

N

S

S N

N S

S

N

N

S

N

S

N

S

S

N

S

N

N

S

S

N

S

N

S

N

N

S

N

SA A

B

U WW'U' U'

CX

Y

X

Z

S

N

S

N

S

N

S

N

Sj

j+1

i i+1

Detail Area C

Magnets Motion (dx,dy)


Vol. 4, No. 6, 2010

1122

Fig.3 Compensation method of unexpected torque. The torque can be canceled and controlled by moment

balance theorem while keeping x and y force independently.

Bxi+1,j(dx,dy)=Bz0・cos(dx/Mw・π/2)・cos(dy/Mw・π/2) (4) Byi+1,j(dx,dy)=Bz0・sin(dx/Mw・π/2)・sin(dy/Mw・π/2+π) (5) Bxi,j+1(dx,dy)=Bz0・sin(dx/Mw・π/2+π)・sin(dy/Mw・π/2) (6) Byi,j+1(dx,dy)=Bz0・cos(dx/Mw・π/2)・cos(dy/Mw・π/2) (7) Bxi+1,j+1(dx,dy)=Bz0・cos(dx/Mw・π/2)・sin(dy/Mw・π/2+π) (8) Byi+1,j+1(dx,dy)=Bz0・sin(dx/Mw・π/2+π)・cos(dy/Mw・π/2) (9) where, i and j coincides with the location of detail area C in Fig. 2 and dx and dy are displacement of stage motion. To explain how to obtain the constant force, y motion is neglected for easy understanding, dy = 0. Provided that the same form of sinusoidal current based on the stage motion is excited into each coil, we get following consequences. fx(dx,0)=Bz0・Ii,j・sin2(dx/Mw・π/2)+Bz0・Ii+1,j・cos2(dx/Mw・π/2) (10) fy(dx,0)=Bz0*Ii,j+1・cos2(dx/Mw・π/2)+Bz0・Ii+1,j+1・sin2(dx/Mw・π/2) (11) Here, fx and fy are a motor force per unit length of active coil, Ii,j and Ii+1,j are input current for x motion and the same value in this case while Ii,j+1 and Ii+1,j+1 should be zero. Finally, Eq. 10 becomes fx(dx,0)=Bz0・Ii,j that is constant as anticipated.

Next, let me explain the method of torque control. As shown in Fig.2, there are four sets of units which are composed of nine magnets and four coils in this example. The electromagnetism in physics mentions that the Lorentz force is orthogonally generated at coils excited in mutual and magnets are repelled back by reaction force consequently. It means that driving points on the moving stage are frequently varied and an unexpected torque is periodic at the condition of uniform current excitation to each coil. Figure 3 shows how to compensate and control the torque. If coils are excited by the same current, the unexpected torque is as following, where from F1 to F4 indicates a Lorentz force at each coil and L1(x), L2(x), L3(y) and L4(y) are a function of stage position(x,y) respectively. Torque= 2・F1・L1(x)+2・F2・L2(x)+2・F3・L3(y)+2・F4・L4(y) (12) Provided that a moment balance theorem is adapted to the torque compensation while keeping total amount of x and y force, the torque can be controlled by using the following

S N S N S NSN S N S N



L1(x) L2(x)

F1 F2

L4(y)

L3(y)F1 F2

F3 F3

F4 F4

X

Y

without torque compensation Fx = 2*F3 + 2*F4 Fy = 2*F1 + 2*F2 Tz = 2*F1*L1(x) + 2*F2*L2(x) + 2*F3*L3(y) + 2*F4*L4(y)




L1(x) L2(x)

F1 F2

L4(y)

L3(y)

F1 F2

F3 F3

F4 F4

X

Y

with torque compensation Fx(y) = 2*F3(y) + 2*F4(y) Fy(x) = 2*F1(x) + 2*F2(x) Tz(x,y) = 2*F1(x)*L1(x) + 2*F2(x)*L2(x) +2*F3(y)*L3(y) + 2*F4(y)*L4(y)

Zero Torque Condition F1(x)*L1(x)=F2(x)*L2(x) F3(y)*L3(y)=F4(y)*L4(y)


Vol. 4, No. 6, 2010

1123

explicit equations: Torque(x,y)=2・F1(x)・L1(x)+2・F2(x)・L2(x)

+2・F3(y)・L3(y)+2・F4(y)・L4(y) (13)

2・F1(x)・L1(x)=2・F2(x)・L2(x)+Tcom/2 2・F3(y)・L3(y)=2・F4(y)・L4(y)+Tcom/2 (14) 2・F1(x)+2・F2(x)=Fxcom 2・F3(y)+2・F4(y)=Fycom

Here, Fxcom, Fycom and Tcom are force and torque command into a system.

If both the back yoke and coil yoke are magnetically saturated, a magnetic leak is emitted and surrounds the planar motor. This means not only that the performance of motor will be strictly limited but also that there will be an electromagnetic influence or coupling against a fine stage mounted on the surface stage. At any rate, it became a de facto standard that isolating and de-coupling mechanically, thermally and electromagnetically between fine stage and coarse stage should be observed in a lithography tool. Therefore, some ideas, for instance the Halbach method of magnet array or optimization of edge effect, may be actually applied. During the development of planar motor, an optimization program of planar motor design has been produced. Figure 4 shows a framework of the program in which there are basically four regions as follows: 1) Input region

a) design specifications : for instance moving mass, maximum acceleration and velocity, stage x and y stroke and electric requirements

b) design parameters : size and pitch of magnet, wire gage and thickness of coil 2) Database region

Several kinds of magnetic FEA(finite element analysis) which parameters are thickness of magnet and magnetic gap between magnet and coil yoke have been implemented beforehand in order to make a matrix of magnetic flux density at active coil location.

Fig. 4 Design program of planar motor that consists of four regions; input, database, calculation and output

region. Regarding a database region, results of magnetic FEA is transformed into a numerical matrix.

Here is Jean-Marc's magnetic f ield estimate, based on finite element analys is. He prov ides tw o functions,Bmin(gap,Zm) and Bmax(gap,Zm) w hich return the strength of the B-field over the center point of the magnet, at thetop and bottom of the air gap, respectively. Both functions take the magnet pitch and gap thickness as parameters.

To be conservative, w e use Bmin as the approximate Peak f ield in the air gap.

This matrix has the gap thicknesss (1st column) andmagnet thickness (2nd column) used in the FEAcalculations.

Maxz and Minz (below ) have the FEA results asmatrices. Each column in those matrices is f or a singlegap size, each row is for a particular magnet thickness.

Mxy

4

8

12

16

20

4

8

12

16

20

mm⋅:=

Maxz

.630

.420

.325

.277

.249

.842

.627

.512

.448

.409

.940

.734

.616

.547

.504

.990

.793

.675

.605

.559

1.018

.827

.710

.638

.592

tesla⋅:= Minz

.629

.408

.289

.212

.158

.838

.597

.439

.329

.249

.933

.693

.521

.396

.303

.981

.745

.568

.435

.335

1.008

.776

.596

.459

.355

tesla⋅:=

vs lspline Mxy Maxz, ( ):= ws lspline Mxy Minz, ( ):= Intermediate results for interplation.

Bmin gap Zm, ( ) interp ws Mxy, Minz, gap

Zm

,

:= These two f unctions return the Bf ield overthe center point of the magnet, at the topand bottom of the air gap.

Bmax gap Zm, ( ) interp vs Mxy, Maxz, gap

Zm

,

:=

Use the minimum value of Bpeakfor safety.Bpeak Bmin Zg Zm, ( )

→:=

Design Specificationsmload 28.5 kg⋅:= Payload mass αreqd 1.6 g⋅:= Acceleration required

Imax 20 amp⋅:= Max coil currentvreqd .50

msec

⋅:= Max velocityVmax 75 volt⋅:= Supply voltage

duty .16:= Duty cycletravel 450 mm⋅:= Stage travel

Xs 430 mm⋅:=Stage dimensions Pitch must be 1/(n*12) of the stage dimension. Since big

magnets have higher field, 1/12 is probably best.Ys Xs:=

Design VariablesPt

43012

mm⋅:= Magnet Pitch. Zm 16 mm⋅:= Magnet thickness

checker 1:= Checker type; 1=normal, 2=nocorners, 3=2 with transversemagnets

arraysize 4:= Size of awg and Zc vectors

i 0 1, arraysize..:= Index variable for vectors andmatrices.awgmin 10:= Wire gages Zcmin 5.5 mm⋅:=

awgmax 25:= Zcmax 10 mm⋅:= Thickness of coils

RESULTSParameters:

awg( )T 10.00 13.75 17.50 21.25 25.00( )=Zm 16.00mm= Magnet thicknessZc( )T 5.50 6.63 7.75 8.88 10.00( ) mm=Pt 35.83 mm= Pitch

Km

13.69

13.70

13.69

13.59

13.48

13.98

13.89

13.88

13.78

13.67

13.89

13.98

13.91

13.80

13.69

13.86

13.86

13.77

13.69

13.58

13.57

13.63

13.57

13.48

13.38

newton

watt= Power

871.24

869.24

871.28

883.92

898.44

834.73

846.03

846.49

859.30

872.99

845.60

835.07

843.73

856.88

870.99

849.49

849.96

860.28

870.40

884.40

886.34

877.86

886.50

897.57

911.75

watt=

Istep

136.88

57.75

24.48

10.50

4.51

121.19

51.94

21.97

9.43

4.05

113.67

47.47

20.28

8.71

3.74

105.88

44.80

19.17

8.20

3.52

102.49

42.92

18.33

7.85

3.37

amp= Vstep

3.40

8.03

18.99

44.85

105.92

3.69

8.72

20.61

48.67

114.93

3.98

9.43

22.27

52.58

124.15

4.29

10.15

23.97

56.62

133.71

4.61

10.90

25.76

60.86

143.72

volt=

Motor Constant Power Consumption

Input Current Input Voltage

MotorDesignEngine

Transition form FEA results to numerical matrix

Magneticflux density

FEA Results

INPUT REGION DATABASE REGION

OUTPUT REGION


Vol. 4, No. 6, 2010

1124

3) Calculation region Details are not shown in Fig. 4. It is just a mathematical calculation of vector and matrix. The range of parameters might be changed once a singular value is obtained.

4) Output region Result of the calculation based on two parameters (row and column) shows motor constant and power consumption, etc. The results proves that a set of a magnet pitch and thickness of 35.8 mm and 16 mm,

wire gage of around 17.5 and magnetic gap of around 7.5 mm is the best solution in this case, then an amplifier or other key design can be started based on this results. 2.2 Air bearing system

Because magnets of planar motor are arranged like a checker board as mentioned in 2.2, multi air pad units can be installed between magnets. An advantage of this concept is that a magnetic attraction force between two yokes is used to get a preload for air bearing to maintain the higher stiffness of levitation. A schematic diagram is shown in Fig. 5 and Fig. 6 that shows a cross section of the surface stage to explain a preload of the system. A set of positive stiffness of air levitation and negative stiffness of magnetic preload makes a system quite stable and is very similar to conventional air bearing that is made up of air levitation and vacuum preload unit.

Fig. 5 Schematic view of air bearing system. Air pads and magnets are aligned every other portion such as a

checker board. a) bottom view of a surface stage, b) one idea of air pad pattern and c) dimensions of air pad.

Fig. 6 Schematic view of preload for air bearing. Because higher stiffness of air bearing is required to obtain a

higher positioning controllability of lithography stage, the balance point at which the air levitation force

corresponds to magnetic attractive force plus gravity force of moving stage is a few micrometers higher than

maximum stiffness point experimentally.

NS

Permanent Magnetsfor planar motor

a) Surface stage(bottom view)

b) Monolithic ceramic body

A

Detail　Area A(air pad)

Air Groove:depth=10～20 µm

35

354

4

44

18

18 1.25

1.25

unit:mm

Depth:12µm4.7

c) Demenstion of air pad

N

S

S

N

Air PadMagnet

Guide Planefor Air Bearing

Armature CoilCoil Yoke

Air levitation forceby Air bearing

Magnetic attractiveforce & Gravity forcefor Air bearing

Movingparts

FixedParts

Z

X,Y

Magnetic attractive force+ Gravity force

Air bearingforce

Air Gap

Fz+

Fz+

gbgp

Stiffnessof air bearing

gp : gap at max. stiffnessgb : gap balancedDesired gap condition: gp = gb + a few µm

//

//

0

Levi

tatio

n Fo

rce


Vol. 4, No. 6, 2010

1125

2.3 Motor cooling system Fine temperature control of sub-deg C is necessary to guarantee a stage performance of

µm or sub-µm level positioning control. Especially, if armature coils are installed inside an air guide block as described in this article, it is necessary to regulate a variation of surface deformation of air guide surface less than 0.5 µm which is almost equal to a temperature rise of the guide surface of 0.1 degC and to a rise of armature coils of 10 degC approximately. Therefore, the amount of heat generated by coils should be conducted to the opposite side of air guide. Motor cooling concept, so called "parallel flow" is shown in Fig. 7. An upper flow of laminar condition makes a laminar boundary layer inside the upper channel isolating the heat transfer from coils to the air guide surface effectively. On the other hand, a lower flow of turbulence condition makes a higher heat conductance inside the lower channel promoted the heat transfer from coils to the bottom of coil yoke. Because a thickness of thermal boundary layer is theoretically decided by flow rate and properties of coolant, a gap of top channel may be larger than the thickness to maintain a boundary layer region(11). The thickness of layer is 1.5 mm and the gap of the channel is 2 mm in this article. To identify a top and bottom flow rate and other critical scales, a model of heat transfer network shown in Fig. 8 was earlier constructed. All considerable heat transfer paths and models, such as heat conduction and convection are included in the network.

Fig. 7 Schematic view of Parallel flow cooling concept. The objective of top flow is to isolate a heat transfer to

an air guide by using thermal boundary layer of laminar flow; otherwise the objective of the bottom flow is to

promote a heat transfer to a coil yoke by using a turbulent flow.

Fig. 8 Model of heat transfer network. This technique is used to solve simultaneous equations in some fields that

can be realized to an equivalent circuit of electricity.

N

S

S

N

Air guide Armature coil

Coil yoke

Z

X,Y

Top flowfor isolation

Bottom flowfor remove

Origin of heat

Laminar boundarylayer

Thin Film

Thickness ofThermal BoundaryLayer

Flow channel

Heat transferisolation flow(Fluorinert cooling)

Heat transferpromotion flow(Water cooling)

Air guideS f

Thin filmArmaturecoil

Q1 Q2

R1 R15Tc1

R2 Tc2

R3 Ttopflow Qf Q5 R16

R4 TpostR5 R6

Rcoil R13QUTc

Rcoil QL R14

Q6Tb1

R8 R7Tbtmflow Qw

R11R9 R10

R17Q7

R12

Q3 R18

TaQ4

T

Qf

Heat flux transferredto the upper flow

Qw

Heat flux transferredto the lower flow

Q1 Q2

Post

Coil Yoke

Cooling can

Remaining heattransferred tothe air

Heat conductedthrough the post

Q3

Q4Ta

Ta

Ta Ta

Ta: Room temperatureHeat flux transferedto the support frame

Q(Heat Input) Heat balance(main equation)

Q=Qf+Qw+Q1+Q2+Q3+Q4


Vol. 4, No. 6, 2010

1126

The model is made up of heat balance equation and some thermal resistance equations expressed by a temperature gradient as followings. Q = Q1 + Q2 + Q3 + Q4 + Qf + Qw (15) dTi = Ri・Qi {i=1,2...Ne}, (16) where Q is an input heat generated by coils, Q1 and Q2 are heat transferred through an upper flow and through a post to air guide surface, Q3 and Q4 are heat transferred through a post to support frame which is not shown and finally Qf and Qw are heat absorbed to upper and lower flow that are illustrated in Fig. 8. Also, dTi, Ri and Qi indicate a temperature gradient, heat resistance and transferred heat between two points in heat transfer network in which Ne is a number of thermal resistance equations in the network respectively. Therefore, each temperature at any points of network can be identified by solving these simultaneous linear equations. Since a top flow should be laminar, cross section of post at the top flow channel is strongly taken into account. After watching around aerospace industries to investigate an appropriate airfoil, the author finally found a NACA64 A218 shape shown in Fig. 9.

Fig. 9 Post shape in a top flow to keep a laminar flow for making a thermal boundary layer.

Fig. 10 FEA of thermal flow dynamics. The model represents a mirror boundary in parallel of flow direction and

a half pitch offset of column of airfoils. An input heat is a constant 20 W per coil which is a worst case and a flow

rate is an upper flow of 1.5 L/min and lower flow of 7.5 L/min. A commercial software "STREAM" is used.

Shape date of NACA64 A218 is followed.

xx100 0 1.25 2.5 5.0 7.5 10 15 20 30 40 50 60 70 80 90 95 100( ):=

yy100 0 2.28 3.2 4.52 5.52 6.36 7.66 8.64 9.88 10.26 9.67 8.4 6.65 4.60 2.34 1.19 0( ):=

rtop100 2.25:= Airfoil shape & Radius of leading edge

This data is indicated in percent of wing length.

lairfoil xreal100:= lairfoil 30.723 mm⋅= Wing length

wmaxairfoil 2 max yreal( )⋅:= wmaxairfoil 6.308 mm⋅= Max wing thickness

Reairfoilvup lairfoil⋅

νfc77:= Reairfoil 3.969 10

3⋅= Reynold' number

Cdairfoil 0.1:= Drag Coefficient

Dairfoil1

2Cdairfoil⋅ ρfc77⋅ vup

2⋅ lairfoil⋅ tgapup⋅:= Dairfoil 5.842 10

5−⋅ N⋅=

Dttlairfoil Npost Dairfoil⋅:= Dttlairfoil 9.464 103−

⋅ N⋅=

Shape Data of NACA64 A218(percent value)

Real post(airfoil) design

Length m

Stream linein flow channel

Post

Wid

th m

air foil

Coil Thermal boundary layerheat input : 20 W at each coilupper flow : 1.5 L/min , fluorinertlower flow : 7.5 L/min , fluorinert

Section AA

A A

2mm

Post


Vol. 4, No. 6, 2010

1127

To conform to how to maintain a thermal boundary layer by using the shape of NACA64 A218 in advance, FEA of thermal flow dynamics was carried out. One of the results is shown in Fig. 10 that seems as if a thermal boundary layer is relatively maintained. Furthermore, this figure shows a worst case scenario of simultaneous excitation of three coils, so a real case of independent switching excitation will be much better than this.

3. Experiment

3.1 Planar motor The test stand to verify a planar motor performance is shown in Fig. 11. To facilitate

testing and reduce development costs, the author employed the "proof of concept" philosophy and therefore, verification of air bearing and motor cooling was separated from this stand. Fig. 11 Test stand for verification of planar motor performance. The moving object of planar motor is attached to

the XY guide with 6 DOF force sensor block

Fig. 12 Moving part and stationary part of planar motor. The coils are wound by round shape wire and attached

to the back yoke. On the other hand, the magnets of Nd-Fe-B are glued to the yoke.

Stage Maintenance Space X-Avis LinearBall Guide with Encoder

Z-AxisRoller Support

Z-AxisGuide Surface

Y-Axis LinearBall Guide

Moving Magnet Stage Fixed Armature Coil Array

Y-Axis Encoder

6 Axis ForceSensor

unit:mmSection BB

Stage Frame

X,Y,θz Stage

200

360

６ dof Force SensorOPFT-1kN-CH-B(Minebea)

670(

coil

arr

ay)

900(

joub

an)

□ □

2000

B

B

430(

mag

net

arra

y)Y

X

Z

Guide Support Beam

Coil Yoke(steel) Armature Coil

□143.3 430 24

430

71.66

43071.66

Permanentmagnet(Nd-Fe-B) Magnet Yoke

(steel)

unit:mm

Moving Part

Stationary Part


Vol. 4, No. 6, 2010

1128

Table 1 Comparison table of magnetic properties of planar motor

The stage is constrained by X, Y linear guide with X and Y encoder and Z rotation guide with 6 DOF integrated force sensor in order to measure X and Y displacement and generated force and torque. Moving part and fixed part of planar motor for test stand is shown in Fig. 12. The moving magnet type motor as early mentioned indicates the moving part is composed of magnets and the field is magnetically limited. Therefore, number of magnets and edge shape of magnet should be experimentally and analytically optimized to avoid a saturation of magnetic flux inside yokes and for a uniform distribution of magnetic field on the magnet array as depicted in Fig. 12. The amplitude of magnetic flux density at air gap and force constant are shown in Table 1 which is compared with design values and FEA results. The difference between numerical design or FEA, and measurement results on force constant may come from the following reasons. a) Armature coil was designed at a minimum tolerance(minimum dimensions) b) Permanent magnet was used at a minimum properties in FEA(minimum B-H curve)

It will be evident to obtain a good coincidence provided that parameters of each element are finely adjusted based on manufacturing results in terms of productivity.

Next, dynamics test of the motor were performed. Figure 13 indicates a response of position feedback in which the stage is controlled by encoders and the result implies a bandwidth of servo is limited around 25 Hz due to a mechanical resonance frequency of 75 Hz which was caused by guide support beam. The target specification of control bandwidth of 50 Hz shown in 2.2.1 will be satisfied if an air bearing system is adapted to the planar motor because the system becomes a monolithic body. Furthermore, as a canceling method defined in Fig. 3 is adapted, a coupling between x/y force and z torque is achieved as less than 10 %. A simple, well used classic controller, PID, was adopted in this test, so the results will be much better when the better and more effective methods that have recently been developed are used, such as a modern controller, a robust or H-infinity controller, a model base Fig. 13 Frequency response of position feedback control with torque cancelling. Horizontal and vertical axis is a

frequency and position to position ratio respectively after fourier transfer from a measurement data of time

domain.

0

101

a) X position command to response

0

b) Y position command to response

100

Bandwidth : ～25Hz

Gai

n dB

-70

+10

+180

-180

Phas

e de

g

0

Mechanicalresonance frequency

Gai

n dB

-70

+10

+180

-180

Phas

e de

g

0

101 100Frequency Hz

101 100

1 100

Frequency Hz

Bandwidth : ～30Hz

10


Vol. 4, No. 6, 2010

1129

controller, etc. Also passive or active vibration compensator can be used for the test; for example, a mass dumper or piezo actuator which is commonly used to decrease amplitude of resonance frequency in order to increase a bandwidth of control. These means are good for practice use, however it was not possible to prove the vibration compensation experimentally because series of test was finished over 10 years ago. 3.2 Air bearing system

Quarter part of full scaled air bearing unit shown in Fig. 14 was made to verify a manufacturability and controllability of flatness of monolithic ceramic block with multi air pad. Figure14 shows a picture of quarter unit under measurement and results of height and flatness of each air pad. Height of each air pad was measured from regulated Jouban and the variation is less than 1.5 µm, so that it is no influence for an air levitation of 5 µm. In addition, slow grindings were repeatedly needed to obtain a desired flatness, however there was no critical difficulty during processing.

Fig. 14 Manufactured quarter model of air bearing. The results of height and flatness is also attached.

Dimensions of air pad groove is same as a pattern shown in Fig.5.

Table 2 Results of air bearing test. Upper table shows the results of individual properties of each air pad, and

lower table shows total properties of five air pads.

a) Individual Air Supply Design +/-(20%) 3σPad No. - #1 #2 #3 #4 #5Air Gap µm 5 - 5 -Supply Pressure MPa 0.455 - - - - - - -Air Flow Rate NL/min 1.75 - 0.42 0.43 0.54 0.71 0.65 0.55 - 0.39Load N 259.0 51.8 296.1 302.1 281.1 267.6 302.5 289.9 12% 45.5Stiffness N/µm 32.5 6.5 29.5 28.0 27.3 24.5 28.8 27.6 -15% 5.8

b) Batch Air Supply for five pads Design +/-(10%) Err[%]Air Gap µm 5 - 3 5 7 3 5 7 5Supply Pressure MPa 0.45 - 4.6Air Flow Rate average NL/min 8.75 - 2.24 3.57 4.93 2.94 4.50 6.02 -

3sigma NL/min - - 0.00 0.03 0.03 0.06 0.03 0.09 -Load average Ｎ 1293.0 129.3 1419.1 1214.0 1045.7 1433.3 1191.8 979.7 -8% 3sigma Ｎ - - 9.11 4.41 5.88 20.87 19.11 21.46 -Stiffness average N/µm 162.5 16.3 113.9 92.5 77.2 131.2 111.4 103.5 -31%

3sigma N/µm - - 3.23 1.76 2.35 3.53 5.88 2.94 -

0.39 0.45

Measurement Average

Measurement

50.45


Vol. 4, No. 6, 2010

1130

Table 2 indicates the results of air bearing experiment. The upper table shows characteristics of individual air pad and the measurement of the air flow rate is three times smaller than design value at large because an effective cross section area is tightened by a steep change of the area. On the other hand, the experimental values of load and stiffness agree well with the design results. The lower table shows the air bearing performance of five pads that the air is simultaneously supplied to the five pads by single tube and experimental results are relatively smaller than design value because the lower height of air pad which means larger air gap makes the performance inferior. The author carried out the experiment as follows. a) Independence of air gap : 3, 5, 7 µm b) Independence of supply pressure : 0.39, 0.45 MPa The results are almost same as anticipated values and it is said that the air gap of maximum stiffness is less than 3 µm. Moreover, it came to light that the gradient of stiffness per unit pressure at 5 µm gap is approximately 300 m and the gradient of stiffness per gap at 0.45 MPa is around -7 N/µm2. A numerical design was done by using nikon's original and reliable software11). 3.3 Motor cooling system It is well known that a heat expansion and fluctuation intensively affect position control of the lithography stage, therefore the temperature rise is restricted less than critical value which comes from the permissible error for the measurement. The test fixture that is manufactured based on the concept of Fig.8 to Fig.10 and the results are shown in Fig. 15 and Fig.16. In this test, the flow rate of the upper and lower flow is 1.5 L/min and 7.5 L/min respectively and the coolant is fluorinert produced by 3M. The heat generation of coil excitation is also constant of 10 W per coil and the thermal expansion coefficient of each material is almost equaled to the real system. Fig. 15 Test stand for verification of cooling system. Inlet and outlet temperature of flows and top and bottom

plate are measured.

Upper FlowInput

Lower FlowInput

RTD103

RTD101

RTD102

RTD104

RTD210

RTD101:Temperature@Lower Flow InletRTD102:Temoerature@Lower Flow OutletRTD103:Tempertaure@Upper Flow InletRTD104:Temperature@Upper Flow OutletRTD210:Temperature@Top plate

FlowDirection

a) Test Stand(full assembly)

C

Section CC

Upper Flow ChannelLower Flow Channel

Armarture Coil

Kapton Film

C

b) Test Stand without Top Plate

Kapton Film betweenUpper Flow & Coils

Armature Coils

Input power at each coil : 10 WTotal power : 10 W x 4 coils = 40 W Flow section is chnaged from circle

to thin film continuously


Vol. 4, No. 6, 2010

1131

Fig. 16 Test results of cooling system. Dynamics is also measured. Time constant is about 20 minutes.

Table 3 Summary of cooling test

To check a possibility of the model simulated prediction, the test results were compared with simulation results shown in Table 3. The results show good mutual relations between experiment and simulation except one thing that a temperature rise of the lower flow is much higher than simulation result. But, it is not easy in fact to explain the rise by the heat removal theorem of heat flow convection, so additional heat may be coming into the system, for example, remaining wire for coil excitation exists in the lower flow region. Since even the on-the-spot survey did not disclose the reason definitely, the author has surmised the problems lie in a failure on the test stand for identification. The thermal modeling seems to be a worthwhile subject to investigate for both accustomed environment and vacuum situation.

4. Conclusion

In this paper, a higher stability of surface stage with Lorentz type planar motor and air levitation was proposed comparing with other type of stages, for instance magnetic levitated stage. After the procedure of each design was introduced concretely, the experimental results of planar motor, air bearing and cooling system were indicated step by step. Regarding the statics of planar motor, the experimental results agree qualitatively and quantitatively with simulation results, however the dynamics shows that the bandwidth of experiment was two times lower than a target specification because of structural difference between production stage of Fig. 1 and test stand of Fig. 11. Next, a performance of air bearing generally agrees with fluid simulations and a stable stiffness of levitation coupling

Temp. rise @ upper flow = 0.14 degC

Temp. rise @ lower flow = 0.69 degC

Room temp.

Time sec

Tem

per

atu

re d

egC

Temp. rise @ top plate = 0.08 degC

Temp. rise @ bottom plate = 0.12 degC

0W 40W

Step response from 0W to 40W at coils

Time Constant


Vol. 4, No. 6, 2010

1132

with magnetic attractive force of planar motor was conformed. Finally, notified errors were not existed in a cooling experiment, however there were some different trends between test and calculation results due to additional paths of heat transfer in the test stand. More accurate modeling will be conducted in near future work in order to obtain a higher certainty of prediction.

The author concludes from the results described above that the surface stage with air bearing system is worth finding a solution for the next generation lithography. A further study regarding vacuum compatibility and possibility of material recycling will be carried out in next phase.

Acknowledgements

The author would like to acknowledge the assistance, support and efforts of Takaharu Miura, and the close collaboration with his colleagues Andrew Hazelton and Mike Binnard as well as the others of the staff of our Nikon Research Center of America, without whom this work would not have been successful.

References

(1) The Magnetic Levitation Technical Committee of the Institute of Electrical Engineers of Japan, Magnetic Suspension Technology,(1993),p.240,CORONA.

(2) J.M. Jansen, C.M.M. van Lierop, Moving-magnet Planar Actuator with Integrated Active Magnetic Bearing, ASPE Conference, (2006), Technical Session VIII.

(3) J.M. Jansen, C.M.M. van Lierop, E.A. Lomonova, and A.J.A. Vandenput, Magnetically Levitated Planar Actuator with Moving Magnets, IEEE Transactions in Industry Applications, Vol.44(2008), No.4, pp.1108-1115.

(4) CM.M. van Lierop, J.W. Jansen, E.A. Lomonova, A.A.H. Damen, P.P.J. van den Bosch, and A.J.A.Vandenput, Commutation of a magnetically levitated Planar Actuator with Moving-Magnets, The International Symposium on Linear Drives for Industry Application, (2007), OS6.2.

(5) Tanaka, K., Development of 6 DOF Magnetic Levitation Stage for Lithography Tool -Background and First Report-, JSPE, Vol.75, No.4(2009), pp.501-508.

(6) Tanaka, K., Development of 6 DOF Magnetic Levitation Stage for Lithography Tool -Proof Of Concept(Second Report)-, JSPE, Vol.75,No.5(2009),pp.605-611.

(7) Sakino S., Akutsu K., Selection of Motor for Ultra Precision, Mechanical Design, Vol.45, No.14(2001), pp.37-41.

(8) Wakui S., Control Embedded in Semiconductor Exposure Apparatus, JSME, Vol.103, No.977(2000), pp.250-253.

(9) Fukagawa Y., Miyadai T., Nakamori M., Improvement of Distortion Measurement in Lithography Tool, JSPE, Vol.71, No.11(2005), pp.1410-1414.

(10) Gao W., Tano M., Kiyono S., Tomita Y., Sasaki T., Precision Positioning of a Sawyer Motor-driven Stage - Proposal of the positioning system and experiments of micro-positioning-, JSPE, Vol.71,No.4(2005),pp.523-527.

(11) Dr.Hermann Schlichting, Boundary-Layer Theory, (1987), p.26, McGraw-Hill. (12) Hasegawa D.., Kouda R., Aoyama T., Arai T., Kobayashi Y., Saotome K., Numerical

Analysis of Gas-Bearings based on the Navier-Storkes Equation, JSPE Spring Conference, (2004), N36.

journal of advanced mechanical design, systems, and ... · journal of advanced mechanical design,...

Documents