Page cover image

DIY RF Sputtering Chamber

*More detailed design explanations, build instruction, and operating procedures are yet to be uploaded*

Preface

At the CMU Hacker Fab, we started building an RF sputtering Chamber in the Fall 2024. After a second iteration, the RF sputtering chamber seems to be reliable enough to share.

The sputtering project was originally initiated to support the development of a reliable DIY CMOS process. For context, the NMOS process as of F24 relies on buying p type wafers which already have a clean gate oxide grown on the, and a polysilicon gate contact deposited on top, providing a fab quality gate stack which is further processed into NMOS integrated circuits (ICs). For CMOS ICs, n type wells are needed within the p type substrate, and theses wells cannot be doped while the gate oxide and polysilicon gate contact are present on the wafer/substrate surface. Thus, to create a CMOS process, the n wells would have to be doped before depositing a gate oxide and gate contact. Thus, to create a CMOS process, the fab needs to be capable of depositing a high quality gate oxide and gate contact, instead of buying wafers with the gate stack already present. The gate oxide and it's interfaces are the most sensitive parts of a transistors, as they are sensitive to small amounts of contamination, and small concentration of thin film and interfacial defects. For example to grow a high quality gate oxide of SiO2, requires a remarkably pure environment since ionic contamination at those temperature is extremely mobile, and can ruin the gate oxide.

After assessing the feasibility of creating high quality gate stacks through various methods, we landed on sputtering as the best tool for our situation; a fab that exists in non clean room. We chose RF sputtering, as it is capable of depositing almost any material, including high K dielectrics. Additionally, the vacuum environment of the sputtering chamber is likely o produce higher purity films than if we attempted to create a high purity furnace for growing oxides.

The initial material chose for the the sputter gate stack were Aluminum oxide (double the dielectric constant of SiO2) as the gate oxide, and aluminum as the gate contact. The aluminum oxide is to be deposited a a reactive process, by flowing in O2 during the sputtering process, and using an Al Target. This process selection allows for the use of a single target, and allows for the deposition of both the gate oxide, and gate contact in a single sputtering run, which helps avoid contamination of the gate oxide, since it is covered with aluminum before venting the chamber, and exposing the wafer to potential contamination.

Si - Al2O3 to Si gate stack:

pros

  • High k dielectric

  • Single target needed if reactive sputtering is used

  • Allows for immediate protection of gate oxide (which is very contamination sensitive)

  • Cheap sputtering target material

cons

  • Of all the metals, Al is particularly sensitive to oxygen contamination, and pre-sputtering/target cleaning. So, our system will have to be able to delivery extremely pure Ar, and a very clean vacuum environment to achieve conductive aluminum,

Goal Specifications

Machine

  • 1E-7 torr base pressure

  • 100 watts 14 MHz RF power with >90% impedance matching

  • Stable plasma down to 5 millitorr Ar

  • .5 - 8" adjustable throw distance

  • >10nm/minute deposition rate for Al at 4" throw distance.

  • Balanced magnetron magnetic field

  • 1 sccm O2 flow accuracy

  • 1 sccm Ar flow accuracy

  • Actuatable substrate shield

  • Air cooled magnetron

  • 2 simultaneous process gasses

  • View port

  • Entire system machinable with only a drill press and band saw.

Thin films

  • Al2O3 measured dielectric constant >6

  • Al2O3 surface roughness <5nm

  • Al resistivity of <3E-6 ohm-cm

Chamber + Integrated Air Cooled Magnetron

Concept Diagrams

Chamber
Air Cooled Magnetron

BOM

$924.87 as of 6/30/2025

Part
#
Link
Price ($)
Role

Aluminum disk (1/2" thick, 9" diameter)

2

72.23

Top Plate and Bottom Plate

Pyrex cylinder with flame polished ends(8.86" (225mm) OD, .275" (7mm) thick, 8" long)

1

225

Chamber walls

Viton "L" gasket (BJLGV-8)

2

155.6

Sealing between cylinder and top/bottom plates

Aluminum plate (.63" thick, 6"x6")

1

4.93

Substrate stage

Aluminum threaded rods (1/4"-20 thread, 8" long, 5 pack)

1

12.5

Support substrate stage

316 Stainless steel wing nuts (1/4"-20 threaded, 5 pack)

2

12.72

Support substrate stage

Aluminum Sheet (6"x24" .032")

1

12.95

Used for magnetron atmosphere side ground box, dark space shield plate, and substrate shutter

Vented cup point screw (18-8 stainless steel, 10-32 threaded 1/2" long, 5 pack)

1

6.33

Supports dark space shield plate, vented to avoid trapped air

Wing nuts (316 stainless steel, 10-32 threaded)

1

5.5

Supports dark space shield plate

Washers (Aluminum, 1.25" ID, 2.25" OD, .16" thick, 5 pack)

1

10.17

Stacked to set dark space shielding distance

N52SH Nedymium disk magnet (1" diamter, 1/8" thick)

1

4.30

Center magnet

N42SH Nedymium disk magnets (1/4" diamter, 1/8" thick)

8

4.8

Outer magnets

Alumina washers (2.5" OD, 1.062" ID, 1/8" thick)

2

25

Fills space between side of target/magnet block, and side walls. Protects viton gasket from metal deposition and plama degredation

Plate (low carbon steel, 1/16" thick, 3"x3")

1

5.69

Cathode plate, pole piece, supports magnets

PEEK Socket head screws (1/4"-20, 3/8" long)

4

23.48

Secures Cathode plate into top plate, compresses gasket

Aluminum socket head screws (4-40 threaded, 1/4" long, 10 pack)

1

12.81

Holds down magnetron atmosphere side ground box

Nylon screws (M4 threaded, 20 mm long, 100 pack)

17.19

Secures fans to ground box without shorting to cathode plate.

KF16 Blanks

4

1.4

Sealing unused KF16 Bulkhead ports

Flared collar thumb screw (1/4"-28 threaded, 18-8 Stainless steel, 1-1/4" long)

1

7.35

Substrate Shutter

Viton O rings (1/4" ID, 12" OD, 1/8" thick, 25 pack)

1

17.44

Substrate shutter sealing

Aluminum nuts (1/4"-28 threaded, 10 pack)

1

14.34

Substrate shutter

Total

$929.17

2" diameter, 1/8" thick, high purity aluminum target not included in BOM

Power Supply

The power supply consists of a 100W Radio transceiver, an antenna tuner, and a switching DC power supply to power the radio. The radio supplies a 14 MHz sine wave at 100 watts into an antenna tuner. The antenna tuner uses selectable inductors and a variable capacitor to tune the impedance of the system, and the output is connected to the magnetron via UHF cable.

The antenna tuner achieves 90-95% impedance matching, delivering 90-95 watts of forward power.

BOM

$1025.88 as of 6/30/2025

Pumping System + Vacuum Gauge

A turbo pump was selected to ensure a base pressure of 1E-7 torr prior to depositions. Any turbo pump + roughing pump system could be used, but the HiCube 300 Eco pumping system was used because we already had it prior to beginning the project. The MPT 200 pressure guage was also used simply becasue we already had one.

BOM

$11,399.86 as of 6/30/2025

Gas Flow

Mass flow controllers that can deliver small amounts precise flow are necessary to achieve ethe desired pressures. The MCFs require a fixed pressure of around 10-20 PSIG on the inlet side to maintain accurate flow, so dual stage regulators are used to control pressure between the MFC and cylinder.

BOM

$4,957.09 as of 6/30/2025

UHP Ar and O2 cylinders not included in BOM

SOP for Al deposition

*Pics and much more detailed procedures yet to be uploaded*

  1. Ensure pump, power supply, and fans are all powered off

  2. Lift off the top plate and place its sideways to avoid contaminating sealing surfaces

  3. Use wing nuts to adjust substrate stage to desired height

  4. Place chip/wafer in center of substrate stage

  5. Place top plate onto cylinder.

  6. Ensure that the substrate shutter will not collide with threaded rods

  7. Power on pumping system, turn on the pump.

  8. Use digital control unit arrow buttons to scroll to parameter 707. Click both arrows simultaneously, then ensure parameter 707 is set to 100% by using arrows to increase and decrease value. Set the value by pressing both arrows simultaneously.

  9. Scroll to parameter 340 to view chamber pressure.

  10. Wait until chamber pressure reaches 5E-7 hPa or lower (this will take several hours, but should be to E-6 range within 15 minutes)

  11. Plug in the power strip to power on all fans and MFC

  12. Open the Ar cylinder top valve.

  13. Adjust the dual stage regulator until the Ar line is at 10-15 PSIG

  14. Set the MFC to 10 sccm.

  15. Scroll to parameter 707, and set pump speed to 25% (aka 250 Hz)

  16. Scroll back to parameter 340 to view pressure, which should now be ~1E-3 hPa

  17. Wait until pump reaches 250 Hz

  18. Set MFC to 90 sccm

  19. Turn on switching DC power supply to power on radio

  20. Turn on radio, and tape handle to keep power on

  21. If plasma has not yet struck, adjust left tuning knob on antenna tuner until it does

  22. Minimize reflected power, should be 10 watts or less at 100 watt output

  23. Reduce to 30 sccm, should result in ~1E-2 hPa pressure

  24. Wait 15 minutes for Al target to be cleaned (pre-sputtering)

  25. Carefully tighten shutter thumb screw to move shutter away from target (avoid lateral torque on screw to prevent any leaks)

  26. Wait the desired deposition time

  27. Turn off radio

  28. Switch off the radio power supply,

  29. Set MFC to 0 sccm

  30. Close Ar cylinder

  31. Stop turbo pump.

  32. Wait until pump speed reaches o Hz

  33. Vent chamber with vent on the side of the turbo pump

  34. Remove top plate, collect chip/wafer.

Testing/Verification

  • Chamber reaches 1.6E-7 hPa base pressure

  • Highly conductive Al films have been sputtered

  • Further characterization of films is actively being pursued

Glass microscope slide with Al sputtered onto it

Future/Pending Work

  • Replae radio with signal generator and amplifier

  • Build our own tuner/matching network (its literally just an inductor and variable capacitor)

  • DIY substrate heater

  • DIY QCM

  • Automation/controls

  • More film characterization, and experimentation with various target materials

Lessons Learned

Many failures occurred over two iterations of power supply and chamber. A more detailed collection and description of these is yet to be uploaded

Contacts

Primary Contacts

  • Jay Kunselman

  • Rahim Malik

Additional Contributors

  • Matthew Moneck

  • Marina Wang

  • Shayaan Gandhi

  • Meadow Webster

  • Rachel Lewis

  • Ayan Ghosh

Misc. Images

Last updated