Applications of permanentmagnet sources and arrays

Francis F. Chen

INER, February 24, 2009

Helicon sources are ICPs with a DC B0
This is a commercial helicon source made by PMT, Inc. and successfully used to etch semiconductor wafers. It required two large and heavy electromagnets and their power supplies. Computer chips are now etched with simpler sources without a DC B-field.

New applications require larger area coverage.

Possible uses of large-area plasma processing

Roll-to-roll plastic sheets Smart windows OLED displays

Solar cells, mass production

Solar cells, advanced

Distributed helicon source: proof of principle
Power scan at z = 7 cm, 5 mT A, 20 G, 13.56 MHz, 2.0 ARGON 1.5 N (10 cm )
-3

P(kW ) 3.0 2.5 2.0 1.5 1.0

1.0

12

0.5

7-tube m=0 array 0.0 0 5 10 R (cm) 15 20 25 30

PROBE

Achieved n > 1.7 x 1012 cm-3, uniform to s3%, but large magnet is required.
F.F. Chen, J.D. Evans, and G.R. Tynan, Plasma Sources Sci. Technol. 10, 236 (2001)

The problem with small magnets
-10

Internal field

0

z (cm)

10

QUARTZ TUBE

7 cm
20

BNC connector

17 mm

ANTENNA
1 cm

5 mm

30

MAGNET WINDING

1 cm

A small solenoid

Field lines diverge too rapidly
13 cm 10 cm

External fieldAnnular permanent
magnets have same problem

PVC PIPE

5 cm

However, the external field can be used

Internal field

External field

Place plasma in the external field, and eject downwards

Note that the stagnation point is very close to the magnet

PM helicons: proof of principle
150 100 50 0 Bz (G) -50 -100 -150 -200 -250 -300 0 5

External field
Calculated Measured

Internal field
10 15 z (cm) 20 25 30

7 6 5 n (1010 cm-3) 4

500W, 1 mTorr

D (cm) 3 2 1 0 -5 0 5 r (cm) 10 15 20
Z2, 40 Z2, 35 Z2, 30 Z2, 21 Z2, 1

The bottom curve is when the tube is INSIDE the magnet

Evolution of a multi-tube PM helicon source
Medusa 1. Antenna design 2. Discharge tube geometry 3. Permanent magnets 4. RF circuitry Medusa 1

Next: construction and testing of Medusa 2

Helicon m = 1 antennas
Only the RH polarized wave is strongly excited

Nagoya Type III antenna: symmetric, so RH wave is driven in both directions.

RH helical antenna: RH wave is driven only in the direction matching the antenna¶s helicity.

This antenna has the highest coupling efficiency

Why we use an m = 0 antenna

A long antenna requires a long tube, and plasma goes to wall before it gets out.

An m = 0 loop antenna can generate plasma near the exit aperture. Note the ³skirt´ that minimizes eddy currents in the flange.

Now we have to design the diameter and length of the tube.

The low-field peak: an essential feature

1.4 B(G) 1.2 1.0 R (ohms) 0.8 0.6 0.4 0.2 0.0 1E+11 100.0 63.1 39.8 25.1 15.8 10.0 L=2", 1mTorr, conducting

Low-field peak

The peak occurs when the backward wave is reflected to interfere constructively with the forward wave.
n (cm-3) 1E+12 1E+13

R is the plasma resistance, which determines the RF power absorbed by the plasma,

Designing the tube geometry

1 [ n w a kZ B

Adjust a, H, and [RF so that n and B are in desired range.

This is done with the HELIC code
D. Arnush, Phys. Plasmas 7, 3042 (2000).

Lc is made very large to simulate injection into a processing chamber.

The code computes the wave fields and the plasma loading resistance Rp vs. n and B

Choose a peak at low B, mid 1012 cm-3 density
1.4 B(G) 1.2 1.0 R (ohms) 0.8 0.6 0.4 0.2 0.0 1E+11 100.0 63.1 39.8 25.1 15.8 10.0 L=2", 1mTorr, conducting

Low-field peak

n (cm-3)

1E+12

1E+13

Typical R(n,B) curves at the low-field peak
2.5 B (G) 2.0 1000 464 215 100 46 22 10 d = 3", H = 2", 13.56MHz H = 3 in. 2.0 H = 2 in. H = 1 in. R (ohms) 1.5 2.5 100G, d = 3", 13.56 MHz

R (ohms)

1.5

1.0

1.0

0.5

0.5

0.0 1E+11

1E+12 n (cm-3)

1E+13

0.0 1E+11 1E+12 n (cm-3) 1E+13

Vary the B-field
3.5 100G, H = 2", 13.56 MHz 3.0 2.5 R (ohms) 2.0 1.5 1.0 0.5 0.0 1E+11 1E+12 n (cm-3) 1E+13 Tube diameter d = 4 in. d = 3 in. d = 2 in. R (ohms) 2.0 1.5 1.0 0.5 0.0 1E+11 2.5 3.0 3.5

Vary the tube length
f = 27.12 MHz f = 13.56 MHz f = 2 MHz

1E+12 n (cm-3)

1E+13

Vary the tube diameter

Vary the RF frequency

Final tube design for 13.56 MHz

Material: Pyrex or quartz With aluminum top

Reason for maximizing Rp: circuit loss Rc

Pin ! Prf

Rp R p  Rc

1000 Prf (W) 1000 500 200 100 Loss Pin (W) 100 Unstable equilibrium Stable equilibrium

Rp

Rc : Pin } Prf

Rp Rc

w Rp
10 1E+11 1000 n0 (cm-3)

No helicon ignition

Rc = 1.0 ;
1E+12 1E+13

Stable equilibria

Pin (W)

R p "" Rc : Pin } Prf

100 Prf (W) 1000 500 200 100 Loss

Rc = 0.1 ;
10 1E+11 n0 (cm-3) 1E+12

1E+13

Magnet design for 60-100 G

Vary the outside diameter

Vary the vertical spacing

Final magnet design
8

NdFeB material, 3´x 5´x1´ thick Bmax = 12 kG

6 4 2 0 -2 -4 -6 -8 -10 -12 -14 -16 -10 -8 -6 -4 -2 0 2 4 6 8 10

300 250 200 B z (G) 150 100 50 0 0 2 4 6 z (in.) 8 10 12 D
r (in.)

0.0 0.52 0.92

RF circuitry

For equal power distribution, the sources are connected in parallel with equal cable lengths. The problem is that the cable lengths, therefore, cannot be short. The length Z2 and the antenna inductance L are the most critical.

Allowable values of C1, C2 in match circuit
1600 1400 1200 1000 C (pF) 800 600 400 200 0 0 0.5 1 L (uH) 1.5 2 2.5 3 C (pF) C1(S) C2(S) 1600 1400 1200 1000 800 600 400 200 0 0 50 Z2 (cm) 100 150 200 C1(S) C2(S)

There is an upper limit to each antenna¶s inductance L.

The range of Z2 can be restrictive for large arrays

C1, C2 for N=8, L = 0.8QH, Z1 = 110 cm, Z2 = 90 cm
(unless varied)

Layout of 8-tube test module, Medusa 2

Staggered configuration

Compact configuration

The spacing is determined from the single-tube density profiles to give 2% uniformity

Side view
Aluminum sheet Adjustable height

Z1 Z2

Probe ports

The source requires only 6´ of vertical space above the process chamber

Medusa 2 in operation at 3 kW CW

Radial profile between tubes at Z2
3.5 3 2.5 cm )
-3

2 1.5 1 0.5 0 -25 -20 -15 -10 -5

n KTe

n (10

11

0 r (cm)

5

10

15

20

25

Density profiles across the chamber
Compact configuration, 3kW Side Langmuir probe
10

Compact 3kW, D=7", 20mTorr

0 3.5´ << 4´ below tubes

8 n (10 cm )

6

11

-3

4
Z1, Z1, Z2, Z2, x= x= x= x= 0 3.5 0 3.5

2

<< 7´ below tubes

0 -8 -6 -4 -2 0 y (in) 2 4 6 8

UCLA

Density profiles across the chamber
Staggered configuration, 3kW Bottom probe array

5 4 cm )
-3

Staggered 3kW, D=7", 20mTorr

x (in.)

3 2 1 0 -8 -6 -4 -2 0 y (in.) 2 4 6

-7 0 7 14

-7 0 7 14´

n (10

11

8

UCLA

An linear array of 15 probes
19.75Ó

0.25Ó 4.0"

0.375Ó

H. Torreblanca, Multitube helicon source with permanent magnets, Thesis, UCLA (2008).

UCLA

Density profiles along the chamber
Staggered configuration, 2kW Bottom probe array

5 4 n (1011 cm-3) 3 2 1 0 -8 -6 -4 -2 0 2 4 6 x (in.) 8 10 12 14 16 Staggered, 2kW , D=7", 20mTorr y (in.) -3.5 0 3.5

Density profiles along the chamber
Compact configuration, 3kW Bottom probe array

10 Compact, 3kW, D=7", 20mTorr y (in) 3.50 3.5

8 n (1011 cm -3)

6

4

2

Data by Humberto Torreblanca, Ph.D. thesis, UCLA, 2008.

0 -8 -6 -4 -2 0 2 4 6 x (in.) 8 10 12 14 16

UCLA

APPLICATION TO LIGHT GASES, LIKE HYDROGEN

Hydrogen RnB scans for 13.56 MHz
0.60 B (G) 0.50 20 40 60 80 H = 1.0 in. conducting 13.56 MHz 0.6 B (G) 0.5 20 40 60 80 H = 1.5 in. conducting 13.56 MHz

0.40 R (ohms)

0.4 R (ohms) 1E+11 1E+12

0.30

0.3

0.20

0.2

0.10

0.1

0.00 1E+10

n (cm-3)

0.0 1E+10

n (cm-3)

1E+11

1E+12

0.6 B (G) 0.5 5 10 15 20 H = 2.0 in. conducting 13.56 MHz

0.6 B (G) 0.5 20 40 60 80 H = 1.5 in. insulating 13.56 MHz

0.4 R (ohms)

0.4 R (ohms)

0.3

0.3

0.2

0.2

0.1

0.1

0.0 1E+10

n (cm-3)

1E+11

1E+12

0.0 1E+10

n (cm-3)

1E+11

1E+12

No stable solution for hydrogen. Here, H is distance from antenna to endplate.

Hydrogen helicons in Medusa 2 tube
14 12 Lower hybrid frequency (MHz) 10 8 6 4 2 0 0 20 40 B-field (G) 60 80 100 n = 1E12 cm-3 13.56 MHz

[LH !
Hydrogen Argon

2 [c

2  ;ci

[ n kB w kz B

The lower hybrid frequency [ LH) is 6.5 times higher for H than for Ar and is not << [(RF). To neglect ion motions, need to have [(RF) >> [(LH). Need to decrease B to have lower [(LH), but low B means bad coupling, like ICPs. Since kB is same if we keep 2´ diam tube, we have to increase [(RF) and change n and kz.

Meaning of the lower hybrid frequency
The exact lower hybrid frequency [LH is given by where ;p is the ion plasma frequency.

1 1 1 }  2 2 [LH [c ;c ; p

The last term is negligible except at very low density, so [LH is proportional to B/˜M. In simple helicons, [ is >> [LH and ;c, so the ions cannot move with the RF. When [LH approaches [RF, the ions will move and contribute to the helicon current. Scime et al. have seen increased ion temperatures when [ ~ [LH, but HELIC does not show any great effect there. At [LH, the ion and electron orbits B to B look like this: The blue line is the ion cyclotron orbit, which has been distorted by the LH wave. The red line is the orbit of the electron guiding-center E x B drift. The cyclotron orbits of the electrons is too small to see.

Hydrogen RnB scans for 27.12 MHz
1.2 B (G) 1.0 10 30 50 70 90 H = 1.0" conducting 1.0 1.2 B (G) 20 40 60 80

0.8 R (ohms)

0.8 R (ohms)

0.6

0.6

0.4

0.4 H = 1.5" conducting 27.12 MHz

0.2

27.12 MHz

0.2

0.0 1E+11

n (cm-3)

1E+12

1E+13

0.0 1E+11

n (cm-3)

1E+12

1E+13

1.2 B (G) 20 40 60 80 100 R (ohms)

1.2 B (G) 1.0 20 40 60 80

1.0

0.8 R (ohms)

0.8

0.6

0.6

0.4 H = 3.0 in. conducting 27.12 MHz

0.4 H = 1.5 in. insulating 27.12 MHz

0.2

0.2

0.0 1E+11

n (cm-3)

1E+12

1E+13

0.0 1E+11

n (cm-3)

1E+12

1E+13

There are stable solutions, but n has to be high, requiring LOTS of power.

Compare hydrogen at 27.12 MHz with argon at 13.56 MHz

to get an idea of how the discharges behave in the standard 2´ diam tube
0.8 0.7 0.6 0.5 R (ohms) R (ohms) 0.4 0.3 0.2 0.1 0.0 1E+11 1.0 0.9 0.8 0.7 R (ohms) R (ohms) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1E+11 B (G) 75 50 25 n (cm-3) 1E+12 1E+13 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1E+11 B (G) 125 100 75 H = 2" Hydrogen, 27.12 MHz B (G) 75 50 25 n (cm-3) 1E+12 1E+13 H = 2" Argon, 13.56 MHz 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1E+11 1.6 1.4 H = 3" Hydrogen, 27.12 MHz B (G) 100 75 50 H = 3" Argon, 13.56 MHz

n (cm-3)

1E+12

1E+13

n (cm-3)

1E+12

1E+13

H is essentially the tube length

How does the power deposition look in normal Ar discharges?
1.0 0.9 0.8 0.7 R (ohms) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1E+11 B (G) 100 80 60 20 40

4

* *
1E+12

Argon, 13.56 MHz

Argon @ 13.56 3 P(z) (arb.) 100G, 1.6E12 40 G, 6.3E11

2

1

n (cm-3)

1E+13

0 -1.00

-0.95

z(m)

-0.90

-0.85

4000 Argon @ 13.56 MHz 100G, 1.6E12 40 G, 6.3E11

0.028 0.024 0.020 P(k) (arb.) 0.016 0.012 0.008 0.004 Hydrogen, 50G, 3E11 @ 27.12 MHz Argon @ 13.56 100G, 1.6E12 40 G, 6.3E11 50G, 3E11

3000

P(r) (arb.)

2000

1000

0 0.000

0.000 0.005 0.010 r (m) 0.015 0.020 0.025 0 25 50 k (m-1) 75 100

Here P(z) and P(r) are the power deposition profiles in z and r, and P(k) is the power spectrum. The cases are at two low-field peaks, and the spectrum is almost a pure mode. The dashed line is the location of the antenna.

Compare similar H and Ar discharges
1400 1200 1000 P(r) (arb.) 800 600 400 0.002 200 0 0.000 0.005 0.010 r (m) 0.015 0.020 0.025 0.000 0 20 40 k(m-1) 60 80 100 P(k) (arb.) 0.006 Hydrogen R = 0.564 Argon R = 0.397 0.010 Argon Hydrogen 0.008

0.004

4 Hydrogen Argon

3 P(z) (arb.)

2

This compares the profiles for argon and hydrogen in the same 2 x 2´ tube and at the same conditions: B = 50G and n = 3 x 1011 cm-3. However, f = 13.56 MHz for argon and 27.12 MHz for hydrogen.

1

0 -1.00 -0.95 z(m) -0.90 -0.85

Power deposition profiles for two very different cases
8000 H (in.), endplate 6000 1.5", conduct. 3.5", insul. 140G, 1.3E12 0.04 0.05 140G, 1.3E12 H (in.), endplate 1.5", conduct. 3.5", insul.

0.03 P(r) 4000 P(k) 0.02 2000

0.01

Both are near density peak, but conducting case has pure mode.

0 0 0.005 0.01 r (m) 0.015 0.02 0.025

0 0 20 40 60 80 k (m-1) 100 120 140

7 6 R = 1.67 ; 5 4 P(z) R = 1.41 ; 3 2 1 0 -1.00 140G, 1.3E12 H (in.), endplate 1.5", conduct. 3.5", insul.

P(r) is dominated by the TG mode and does not vary much. P(z) peaks near the antenna (dashed line in each case). High P near endplate is not good, since plasma created there is lost fast. The k-spectrum is pure for H = 1.5´ but has other modes for H = 3.5´, as seen by the wiggles in the RnB curve on the last page.

-0.95

z (m)

-0.90

-0.85

-0.80

Comparison of waves in 1.5 in. and 3 in. long tubes
7 6 5 140G, 1.3E12, conducting |Ez|(z) 4 P(z) 3 140G, 1.4E12, conducting 2 R = 0.87; 1 0 -1.00 -0.95 z (m) -0.90 -0.85 -0.80 1 0 -1.00 -0.95 z (m) -0.90 -0.85 -0.80 140G, 1.4E12, conducting R = 1.67; 4 3 2 H = 1.5 in. H = 3 in. 6 5 R = 1.67; 140G, 1.3E12, conducting H = 1.5" H = 3"

R = 0.87;

The short tube has higher P(z), but it is high near the endplate. The electric field |Ez|, however, fits properly , whereas it is too short for the 3´ tube. The maximum of Ez at the endplate causes strong reflection, which gives a higher low-field peak. Thus, the short tube is better even though a lot of useless ionization occurs near the endplate. This shows that computing Ez may be the best way to fit the tube length to the halfwavelength of the helicon wave and optimize the loading.

Comparison of 3 optimized systems of different diameters
For hydrogen at 27.12 MHz

Tube: 2´ diam, 1.5´ high Magnet: 3 x 5´, 2´ high

Tube: 3´ diam, 2´ high Magnet: 4 x 6´, 2´ high

Tube: 6´ diam, 3´ high Magnet: 7 x 10´, 4´ high

Note: antenna inductance has to be adjusted

APPLICATION TO SPACECRAFT THRUSTERS

A Hall-effect thruster
It requires an electron neutralizer

Generation of a ³double layer´

B n ¨ r0 ¸ ! !© ¹ B0 n0 ª r º

2

ne ! n0 e L , where L | -eV/KTe
The Bohm velocity is reached when L = ½, and sheath forms

n ! e 1/ 2 , and thus n0

r ! e1/4 ! 1.28 r0

Potential jump observed by Charles et al.

B-field in Boswell¶s helicon machine

Medusa source adapted to VASIMR

The optimized 9-cm diam source is shown with dimensions in cm, together with a NdFeB magnet designed for 400G at the antenna. D is the distance from the midplane of the magnet to the midplane of the antenna. The magnet is made in two pieces supported by a non-ferrous metal plate. The B-field can be adjusted by changing D either by hand or remotely with a motor.

A stronger B-field for higher density

Layout of magnet and tube for 600G operation, showing a gas feed line and a DC bias supply.

A small diam source with for testing high-field operation

A 5-cm diam helicon tube and a 600-G magnet designed for a small overall system diameter.

Conclusion on spacecraft thrusters

‡ ‡ ‡ ‡

³Ambipolar´ sources can eject ions with automatic space-charge neutralization. Helicon sources can generate ions efficiently. Permanent magnets can reduce the complexity of helicon sources. However, for the fields and densities considered for the VASIMR project, the magnet may be too large to be practical.